G2P (BERT, byT5)

.gitattributes

@@ -33,3 +33,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+BERT/bene-ges/en_g2p_cmu_bert_large/en_g2p.nemo filter=lfs diff=lfs merge=lfs -text
+BERT/bene-ges/ru_g2p_ipa_bert_large/g2p_correct_vocab.txt filter=lfs diff=lfs merge=lfs -text
+BERT/bene-ges/ru_g2p_ipa_bert_large/ru_g2p.nemo filter=lfs diff=lfs merge=lfs -text

BERT/bene-ges/en_g2p_cmu_bert_large/.gitattributes

@@ -0,0 +1,35 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+en_g2p.nemo filter=lfs diff=lfs merge=lfs -text

BERT/bene-ges/en_g2p_cmu_bert_large/README.md

@@ -0,0 +1,71 @@
+---
+license: cc-by-4.0
+language:
+- en
+library_name: nemo
+pipeline_tag: token-classification
+tags:
+- G2P
+- Grapheme-to-Phoneme
+---
+
+# English G2P token classification model
+
+This is a non-autoregressive model for English grapheme-to-phoneme (G2P) conversion based on BERT architecture. It predicts phonemes in CMU format. 
+Initial data was built using CMUdict v0.07 
+
+
+## Intended uses & limitations
+
+The input is expected to contain english words consisting of latin letters and apostrophe, all letters separated by space.
+
+### How to use
+
+Install NeMo.
+
+Download en_g2p.nemo (this model)
+```bash
+git lfs install
+git clone https://huggingface.co/bene-ges/en_g2p_cmu_bert_large
+```
+
+Run
+
+```bash
+python ${NEMO_ROOT}/examples/nlp/text_normalization_as_tagging/normalization_as_tagging_infer.py \
+  pretrained_model=en_g2p_cmu_bert_large/en_g2p.nemo \
+  inference.from_file=input.txt \
+  inference.out_file=output.txt \
+  model.max_sequence_len=64 \
+  inference.batch_size=128 \
+  lang=en
+```
+
+Example of input file:
+```
+g e f f e r t
+p r o s c r i b e d
+p r o m i n e n t l y
+j o c e l y n
+m a r c e c a ' s
+s t a n k o w s k i
+m u f f l e
+```
+
+Example of output file:
+```
+G EH1  F  ER0 T	               g e f f e r t           G EH1 <DELETE> F <DELETE> ER0 T   G EH1 <DELETE> F <DELETE> ER0 T   PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN
+P R OW0 S K R AY1 B  D         p r o s c r i b e d	   P R OW0 S K R AY1 B <DELETE> D    P R OW0 S K R AY1 B <DELETE> D    PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN
+P R AA1 M AH0 N AH0 N T L IY0  p r o m i n e n t l y   P R AA1 M AH0 N AH0 N T L IY0     P R AA1 M AH0 N AH0 N T L IY0     PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN
+JH AO1 S  L IH0 N              j o c e l y n           JH AO1 S <DELETE> L IH0 N         JH AO1 S <DELETE> L IH0 N         PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN
+M AA0 R S EH1 K AH0  Z         m a r c e c a ' s       M AA0 R S EH1 K AH0 <DELETE> Z	 M AA0 R S EH1 K AH0 <DELETE> Z    PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN
+S T AH0 NG K AO1 F S K IY0     s t a n k o w s k i     S T AH0 NG K AO1 F S K IY0        S T AH0 NG K AO1 F S K IY0        PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN
+M AH1  F AH0L                  m u f f l e	           M AH1 <DELETE> F AH0_L <DELETE>   M AH1 <DELETE> F AH0_L <DELETE>   PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN
+```
+
+Note that the correct output tags are in the **third** column, input is in the second column.
+Tags correspond to input letters in a one-to-one fashion. If you remove `<DELETE>` tag, and replace `_` with space, you should get CMU-like transcription.
+
+### How to use for TTS
+See this [script](https://github.com/bene-ges/nemo_compatible/blob/main/scripts/tts/tts_en_infer_from_cmu_phonemes.py) to run TTS directly from CMU phonemes.
+

BERT/bene-ges/en_g2p_cmu_bert_large/en_g2p.nemo

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98a4b107e0ccfdb863c4db26a928768c3651df5a4aa1de931e4d4335f23eedb9
+size 439398400

BERT/bene-ges/en_g2p_cmu_bert_large/source.txt

@@ -0,0 +1 @@
+https://huggingface.co/bene-ges/en_g2p_cmu_bert_large

BERT/bene-ges/ru_g2p_ipa_bert_large/.gitattributes

@@ -0,0 +1,36 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+ru_g2p.nemo filter=lfs diff=lfs merge=lfs -text
+g2p_correct_vocab.txt filter=lfs diff=lfs merge=lfs -text

BERT/bene-ges/ru_g2p_ipa_bert_large/README.md

@@ -0,0 +1,74 @@
+---
+license: cc-by-4.0
+language:
+- ru
+library_name: nemo
+pipeline_tag: token-classification
+tags:
+- G2P
+- Grapheme-to-Phoneme
+---
+
+# Russian G2P token classification model
+
+This is a non-autoregressive model for Russian grapheme-to-phoneme (G2P) conversion based on BERT architecture. It predicts phonemes in IPA format. 
+Initial data was built using Wiktionary json from https://kaikki.org/dictionary/Russian/index.html 
+
+
+## Intended uses & limitations
+
+The input is expected to consist of cyrillic letters separated by space. Real space should be replaced to underscore(_).
+Note that the model was trained on single words and some short phrases.
+Though it can accept longer phrases its accuracy may degrade on them.
+
+### How to use
+
+Install NeMo.
+
+Download ru_g2p.nemo (this model)
+```bash
+git lfs install
+git clone https://huggingface.co/bene-ges/ru_g2p_ipa_bert_large
+```
+
+Run
+
+```bash
+python ${NEMO_ROOT}/examples/nlp/text_normalization_as_tagging/normalization_as_tagging_infer.py \
+  pretrained_model=ru_g2p_ipa_bert_large/ru_g2p.nemo \
+  inference.from_file=input.txt \
+  inference.out_file=output.txt \
+  model.max_sequence_len=512 \
+  inference.batch_size=128 \
+  lang=ru
+```
+
+Example of input file:
+```
+и с х о д
+т р а н с н е п т у н о в ы х
+т е л я т н и к о в с к о е
+ц а р с к о г о
+к р о с х о ф
+г а н с - ю р г е н
+д а р д а н е л л
+```
+
+Example of output file:
+```
+ɪ s x 'o t                          и с х о д                       ɪ s x 'o t                         ɪ s x 'o t                           PLAIN PLAIN PLAIN PLAIN PLAIN
+t r a nʲ sʲ nʲ ɪ p t 'u n ə v ɨ x   т р а н с н е п т у н о в ы х   t r a nʲ sʲ nʲ ɪ p t 'u n ə v ɨ x   t r a nʲ sʲ nʲ ɪ p t 'u n ə v ɨ x    PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN
+tʲ ɪ lʲ 'æ tʲ nʲ ɪ k ə f s k ə jə   т е л я т н и к о в с к о е     tʲ ɪ lʲ 'æ tʲ nʲ ɪ k ə f s k ə jə   tʲ ɪ lʲ 'æ tʲ nʲ ɪ k ə f s k ə jə    PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN
+t~s 'a r s k ə v ə                  ц а р с к о г о                 t~s 'a r s k ə v ə                 t~s 'a r s k ə v ə                  PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN
+k r ɐ s x 'o f                      к р о с х о ф                   k r ɐ s x 'o f                     k r ɐ s x 'o f                      PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN
+ɡ a n s 'ju r ɡʲ ɪ n                г а н с - ю р г е н             ɡ a n s _ 'ju r ɡʲ ɪ n              ɡ a n s _ 'ju r ɡʲ ɪ n              PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN
+d ə r d ɐ n 'ɛ ɫ                    д а р д а н е л л               d ə r d ɐ n 'ɛ ɫ <DELETE>          d ə r d ɐ n 'ɛ ɫ <DELETE>            PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN PLAIN
+```
+
+Note that the correct output tags are in the **third** column, input is in the second column.
+Tags correspond to input letters in a one-to-one fashion. If you remove `<DELETE>` tag, `+`, `~`, and spaces, you should get IPA-like transcription.
+The model does not predict secondary stress. The primary stress is put directly before the stressed vowel. In some cases stress can be missing.
+
+### How to use for TTS
+
+See example of inference pipeline for G2P + FastPitch + HifiGAN in this [notebook](https://github.com/bene-ges/nemo_compatible/blob/main/notebooks/Russian_TTS_with_IPA_G2P_FastPitch_and_HifiGAN.ipynb).

BERT/bene-ges/ru_g2p_ipa_bert_large/g2p_correct_vocab.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6ea7cf39bdf8b6d5d46c4c5b6511170bc08a5d46c8c5b03fb8ef775afce75a1c
+size 33813179

BERT/bene-ges/ru_g2p_ipa_bert_large/ru_g2p.nemo

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88c28441b02451a5ea75eb293cd92ec3903590d61685ae2d3fb1dcef56c7854f
+size 719523840

BERT/bene-ges/ru_g2p_ipa_bert_large/source.txt

@@ -0,0 +1 @@
+https://huggingface.co/bene-ges/ru_g2p_ipa_bert_large

byT5/Charsiu/CharsiuG2P.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:070970de2794f4efc9345b319398dcbe32fb1b799d8d65ec4b67407fba7443a2
+size 395041832

byT5/Charsiu/CharsiuG2P2 [fred33iferu] +3.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad4c28cc150a97c6ce554e8a57b5bfcd7df95b23b79cb28e93b75ee99b93e991
+size 260025081

byT5/Charsiu/g2p_multilingual_byT5_small/.gitattributes

@@ -0,0 +1,27 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

byT5/Charsiu/g2p_multilingual_byT5_small/config.json

@@ -0,0 +1,29 @@
+{
+  "_name_or_path": "google/byt5-small",
+  "architectures": [
+    "T5ForConditionalGeneration"
+  ],
+  "d_ff": 3584,
+  "d_kv": 64,
+  "d_model": 1472,
+  "decoder_start_token_id": 0,
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-gelu",
+  "gradient_checkpointing": false,
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "num_decoder_layers": 4,
+  "num_heads": 6,
+  "num_layers": 12,
+  "pad_token_id": 0,
+  "relative_attention_num_buckets": 32,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "ByT5Tokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.16.2",
+  "use_cache": true,
+  "vocab_size": 384
+}

byT5/Charsiu/g2p_multilingual_byT5_small/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4ffba3b176a923b4c0d74b9eac65500b48f870cc65d0dee828392b8cd2da777
+size 1198627501

byT5/Charsiu/g2p_multilingual_byT5_small/source.txt

@@ -0,0 +1 @@
+https://huggingface.co/charsiu/g2p_multilingual_byT5_small

byT5/Charsiu/g2p_multilingual_byT5_small/trainer_state.json

@@ -0,0 +1,1096 @@
+{
+  "best_metric": 0.15353840099691743,
+  "best_model_checkpoint": "/scratch/lingjzhu_root/lingjzhu1/lingjzhu/g2p/byt5_small_finetuned_from_pretrained/checkpoint-5000",
+  "epoch": 9.646302250803858,
+  "global_step": 135000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003,
+      "loss": 1.3832,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002999616623572683,
+      "loss": 0.2586,
+      "step": 2000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029984666902607135,
+      "loss": 0.1824,
+      "step": 3000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002996550787873857,
+      "loss": 0.152,
+      "step": 4000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002993869895761197,
+      "loss": 0.1328,
+      "step": 5000
+    },
+    {
+      "epoch": 0.36,
+      "eval_cer": 0.15353840099691743,
+      "eval_loss": 0.16419939696788788,
+      "eval_runtime": 90.9116,
+      "eval_samples_per_second": 54.449,
+      "eval_steps_per_second": 0.429,
+      "eval_wer": 0.40606060606060607,
+      "step": 5000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000299042538431052,
+      "loss": 0.1171,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029862190142478177,
+      "loss": 0.1067,
+      "step": 7000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029812529357372587,
+      "loss": 0.0979,
+      "step": 8000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029755296872820933,
+      "loss": 0.0911,
+      "step": 9000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002969052194427048,
+      "loss": 0.0857,
+      "step": 10000
+    },
+    {
+      "epoch": 0.71,
+      "eval_cer": 0.1219912113858464,
+      "eval_loss": 0.10491432994604111,
+      "eval_runtime": 90.8023,
+      "eval_samples_per_second": 54.514,
+      "eval_steps_per_second": 0.43,
+      "eval_wer": 0.3408080808080808,
+      "step": 10000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002961823768262882,
+      "loss": 0.082,
+      "step": 11000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002953848103733858,
+      "loss": 0.0774,
+      "step": 12000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029451292777490066,
+      "loss": 0.0745,
+      "step": 13000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002935671747098137,
+      "loss": 0.0708,
+      "step": 14000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029254803461736643,
+      "loss": 0.0673,
+      "step": 15000
+    },
+    {
+      "epoch": 1.07,
+      "eval_cer": 0.11350866837192015,
+      "eval_loss": 0.08455779403448105,
+      "eval_runtime": 91.1047,
+      "eval_samples_per_second": 54.333,
+      "eval_steps_per_second": 0.428,
+      "eval_wer": 0.31757575757575757,
+      "step": 15000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00029145602844994243,
+      "loss": 0.065,
+      "step": 16000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002902917144067724,
+      "loss": 0.0625,
+      "step": 17000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028905568764860047,
+      "loss": 0.0611,
+      "step": 18000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028774857999345685,
+      "loss": 0.0599,
+      "step": 19000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002863710595936922,
+      "loss": 0.0585,
+      "step": 20000
+    },
+    {
+      "epoch": 1.43,
+      "eval_cer": 0.10635972103801841,
+      "eval_loss": 0.0741606280207634,
+      "eval_runtime": 90.9345,
+      "eval_samples_per_second": 54.435,
+      "eval_steps_per_second": 0.429,
+      "eval_wer": 0.29757575757575755,
+      "step": 20000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002849238305944389,
+      "loss": 0.057,
+      "step": 21000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00028340763277367477,
+      "loss": 0.0557,
+      "step": 22000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002818232411640713,
+      "loss": 0.0548,
+      "step": 23000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00028017146565682144,
+      "loss": 0.0538,
+      "step": 24000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00027845315058764886,
+      "loss": 0.0526,
+      "step": 25000
+    },
+    {
+      "epoch": 1.79,
+      "eval_cer": 0.10283990293172428,
+      "eval_loss": 0.06704817712306976,
+      "eval_runtime": 91.2941,
+      "eval_samples_per_second": 54.22,
+      "eval_steps_per_second": 0.427,
+      "eval_wer": 0.2898989898989899,
+      "step": 25000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00027666917430520975,
+      "loss": 0.0519,
+      "step": 26000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00027482044872210895,
+      "loss": 0.0506,
+      "step": 27000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002729079188487587,
+      "loss": 0.0498,
+      "step": 28000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00027093256231031885,
+      "loss": 0.0468,
+      "step": 29000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00026889538884696597,
+      "loss": 0.0463,
+      "step": 30000
+    },
+    {
+      "epoch": 2.14,
+      "eval_cer": 0.10310224962287663,
+      "eval_loss": 0.0634964108467102,
+      "eval_runtime": 91.3121,
+      "eval_samples_per_second": 54.21,
+      "eval_steps_per_second": 0.427,
+      "eval_wer": 0.2896969696969697,
+      "step": 30000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002667974397977457,
+      "loss": 0.0461,
+      "step": 31000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002646397875682729,
+      "loss": 0.0454,
+      "step": 32000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00026242353508255185,
+      "loss": 0.0448,
+      "step": 33000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002601498152191957,
+      "loss": 0.0446,
+      "step": 34000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002578197902323352,
+      "loss": 0.0443,
+      "step": 35000
+    },
+    {
+      "epoch": 2.5,
+      "eval_cer": 0.1002382982444634,
+      "eval_loss": 0.06088022515177727,
+      "eval_runtime": 91.0524,
+      "eval_samples_per_second": 54.364,
+      "eval_steps_per_second": 0.428,
+      "eval_wer": 0.282020202020202,
+      "step": 35000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00025543465115751026,
+      "loss": 0.0434,
+      "step": 36000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0002529956172028505,
+      "loss": 0.0431,
+      "step": 37000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002505039351258541,
+      "loss": 0.0426,
+      "step": 38000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002479608785960846,
+      "loss": 0.0417,
+      "step": 39000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0002453677475441111,
+      "loss": 0.0415,
+      "step": 40000
+    },
+    {
+      "epoch": 2.86,
+      "eval_cer": 0.09859863142476115,
+      "eval_loss": 0.05760909989476204,
+      "eval_runtime": 91.2335,
+      "eval_samples_per_second": 54.256,
+      "eval_steps_per_second": 0.427,
+      "eval_wer": 0.27494949494949494,
+      "step": 40000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00024272586749702474,
+      "loss": 0.0411,
+      "step": 41000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002400365889008706,
+      "loss": 0.0408,
+      "step": 42000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023730128643034235,
+      "loss": 0.0377,
+      "step": 43000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00023452135828609167,
+      "loss": 0.0381,
+      "step": 44000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0002316982254800121,
+      "loss": 0.0379,
+      "step": 45000
+    },
+    {
+      "epoch": 3.22,
+      "eval_cer": 0.09984477820773485,
+      "eval_loss": 0.056990817189216614,
+      "eval_runtime": 91.2394,
+      "eval_samples_per_second": 54.253,
+      "eval_steps_per_second": 0.427,
+      "eval_wer": 0.27595959595959596,
+      "step": 45000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00022883333110886237,
+      "loss": 0.0375,
+      "step": 46000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00022592813961660067,
+      "loss": 0.0372,
+      "step": 47000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00022298413604580696,
+      "loss": 0.037,
+      "step": 48000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00022000282527857588,
+      "loss": 0.0367,
+      "step": 49000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0002169857312672683,
+      "loss": 0.0366,
+      "step": 50000
+    },
+    {
+      "epoch": 3.57,
+      "eval_cer": 0.09800835136966835,
+      "eval_loss": 0.05443936586380005,
+      "eval_runtime": 91.3792,
+      "eval_samples_per_second": 54.17,
+      "eval_steps_per_second": 0.427,
+      "eval_wer": 0.27434343434343433,
+      "step": 50000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00021393439625551483,
+      "loss": 0.0365,
+      "step": 51000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00021085037998986924,
+      "loss": 0.036,
+      "step": 52000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020773525892251514,
+      "loss": 0.0361,
+      "step": 53000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00020459062540543316,
+      "loss": 0.0358,
+      "step": 54000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00020141808687644067,
+      "loss": 0.0355,
+      "step": 55000
+    },
+    {
+      "epoch": 3.93,
+      "eval_cer": 0.09678406681095734,
+      "eval_loss": 0.052644312381744385,
+      "eval_runtime": 91.3207,
+      "eval_samples_per_second": 54.205,
+      "eval_steps_per_second": 0.427,
+      "eval_wer": 0.27090909090909093,
+      "step": 55000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019821926503751995,
+      "loss": 0.0351,
+      "step": 56000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019499579502585537,
+      "loss": 0.0327,
+      "step": 57000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019174932457800242,
+      "loss": 0.0326,
+      "step": 58000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001884815131876167,
+      "loss": 0.0327,
+      "step": 59000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018519403125717278,
+      "loss": 0.0324,
+      "step": 60000
+    },
+    {
+      "epoch": 4.29,
+      "eval_cer": 0.09630309787717803,
+      "eval_loss": 0.05228239297866821,
+      "eval_runtime": 91.3618,
+      "eval_samples_per_second": 54.18,
+      "eval_steps_per_second": 0.427,
+      "eval_wer": 0.26686868686868687,
+      "step": 60000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018188855924410722,
+      "loss": 0.0326,
+      "step": 61000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00017856678680182127,
+      "loss": 0.0323,
+      "step": 62000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001752304119159834,
+      "loss": 0.0322,
+      "step": 63000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017188114003657205,
+      "loss": 0.0323,
+      "step": 64000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00016852068320610358,
+      "loss": 0.0316,
+      "step": 65000
+    },
+    {
+      "epoch": 4.64,
+      "eval_cer": 0.09446667103911152,
+      "eval_loss": 0.050846464931964874,
+      "eval_runtime": 91.2629,
+      "eval_samples_per_second": 54.239,
+      "eval_steps_per_second": 0.427,
+      "eval_wer": 0.2638383838383838,
+      "step": 65000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00016515075918448972,
+      "loss": 0.0318,
+      "step": 66000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016177309057097285,
+      "loss": 0.0315,
+      "step": 67000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00015838940392358722,
+      "loss": 0.0315,
+      "step": 68000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00015500142887659688,
+      "loss": 0.0313,
+      "step": 69000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00015161089725636095,
+      "loss": 0.0313,
+      "step": 70000
+    },
+    {
+      "epoch": 5.0,
+      "eval_cer": 0.09420432434795917,
+      "eval_loss": 0.048644062131643295,
+      "eval_runtime": 91.4526,
+      "eval_samples_per_second": 54.126,
+      "eval_steps_per_second": 0.426,
+      "eval_wer": 0.26141414141414143,
+      "step": 70000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00014821954219607845,
+      "loss": 0.0289,
+      "step": 71000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0001448290972498651,
+      "loss": 0.029,
+      "step": 72000
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00014144129550661485,
+      "loss": 0.0288,
+      "step": 73000
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0001380578687040995,
+      "loss": 0.0289,
+      "step": 74000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00013468054634375843,
+      "loss": 0.0292,
+      "step": 75000
+    },
+    {
+      "epoch": 5.36,
+      "eval_cer": 0.0939638398810695,
+      "eval_loss": 0.049041230231523514,
+      "eval_runtime": 91.4249,
+      "eval_samples_per_second": 54.143,
+      "eval_steps_per_second": 0.427,
+      "eval_wer": 0.258989898989899,
+      "step": 75000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00013131105480663235,
+      "loss": 0.0289,
+      "step": 76000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.000127951116470891,
+      "loss": 0.0288,
+      "step": 77000
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00012460244883140783,
+      "loss": 0.0287,
+      "step": 78000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0001212667636218309,
+      "loss": 0.0284,
+      "step": 79000
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00011794576593959775,
+      "loss": 0.0283,
+      "step": 80000
+    },
+    {
+      "epoch": 5.72,
+      "eval_cer": 0.0939419776568068,
+      "eval_loss": 0.04810198023915291,
+      "eval_runtime": 91.3828,
+      "eval_samples_per_second": 54.168,
+      "eval_steps_per_second": 0.427,
+      "eval_wer": 0.2604040404040404,
+      "step": 80000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00011464115337434394,
+      "loss": 0.029,
+      "step": 81000
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00011135461514014796,
+      "loss": 0.0289,
+      "step": 82000
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00010808783121205837,
+      "loss": 0.0285,
+      "step": 83000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00010484247146734352,
+      "loss": 0.029,
+      "step": 84000
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00010162019483190237,
+      "loss": 0.0282,
+      "step": 85000
+    },
+    {
+      "epoch": 6.07,
+      "eval_cer": 0.09354845762007827,
+      "eval_loss": 0.048511628061532974,
+      "eval_runtime": 93.8707,
+      "eval_samples_per_second": 52.732,
+      "eval_steps_per_second": 0.415,
+      "eval_wer": 0.25696969696969696,
+      "step": 85000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 9.842264843227404e-05,
+      "loss": 0.0282,
+      "step": 86000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 9.52514667536784e-05,
+      "loss": 0.0281,
+      "step": 87000
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 9.210827080451842e-05,
+      "loss": 0.0281,
+      "step": 88000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 8.899466728777203e-05,
+      "loss": 0.0279,
+      "step": 89000
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 8.591224777969557e-05,
+      "loss": 0.028,
+      "step": 90000
+    },
+    {
+      "epoch": 6.43,
+      "eval_cer": 0.09381080431123062,
+      "eval_loss": 0.047929223626852036,
+      "eval_runtime": 93.5944,
+      "eval_samples_per_second": 52.888,
+      "eval_steps_per_second": 0.417,
+      "eval_wer": 0.2608080808080808,
+      "step": 90000
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 8.286258791626041e-05,
+      "loss": 0.028,
+      "step": 91000
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.984724658773716e-05,
+      "loss": 0.0276,
+      "step": 92000
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.686776514184009e-05,
+      "loss": 0.0275,
+      "step": 93000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.392566659583846e-05,
+      "loss": 0.0274,
+      "step": 94000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 7.102245485803813e-05,
+      "loss": 0.0276,
+      "step": 95000
+    },
+    {
+      "epoch": 6.79,
+      "eval_cer": 0.09346100872302748,
+      "eval_loss": 0.04663492366671562,
+      "eval_runtime": 93.7458,
+      "eval_samples_per_second": 52.802,
+      "eval_steps_per_second": 0.416,
+      "eval_wer": 0.25737373737373737,
+      "step": 95000
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.81596139590308e-05,
+      "loss": 0.0273,
+      "step": 96000
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.533860729310434e-05,
+      "loss": 0.0275,
+      "step": 97000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.256087687020127e-05,
+      "loss": 0.0271,
+      "step": 98000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.98278425788092e-05,
+      "loss": 0.0256,
+      "step": 99000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.71409014601578e-05,
+      "loss": 0.0257,
+      "step": 100000
+    },
+    {
+      "epoch": 7.15,
+      "eval_cer": 0.09276141754662119,
+      "eval_loss": 0.04688685014843941,
+      "eval_runtime": 93.8485,
+      "eval_samples_per_second": 52.745,
+      "eval_steps_per_second": 0.416,
+      "eval_wer": 0.25696969696969696,
+      "step": 100000
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.4501426994095876e-05,
+      "loss": 0.0261,
+      "step": 101000
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.191076839701103e-05,
+      "loss": 0.0262,
+      "step": 102000
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.9370249932153075e-05,
+      "loss": 0.0261,
+      "step": 103000
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.6881170232712164e-05,
+      "loss": 0.0259,
+      "step": 104000
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.444480163799822e-05,
+      "loss": 0.0261,
+      "step": 105000
+    },
+    {
+      "epoch": 7.5,
+      "eval_cer": 0.09184320412758794,
+      "eval_loss": 0.04592433199286461,
+      "eval_runtime": 93.414,
+      "eval_samples_per_second": 52.99,
+      "eval_steps_per_second": 0.417,
+      "eval_wer": 0.25515151515151513,
+      "step": 105000
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.2062389543061265e-05,
+      "loss": 0.0257,
+      "step": 106000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.9735151762084384e-05,
+      "loss": 0.0258,
+      "step": 107000
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.746427790587557e-05,
+      "loss": 0.0258,
+      "step": 108000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.525092877377602e-05,
+      "loss": 0.0258,
+      "step": 109000
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.309623576029597e-05,
+      "loss": 0.0258,
+      "step": 110000
+    },
+    {
+      "epoch": 7.86,
+      "eval_cer": 0.09195251524890143,
+      "eval_loss": 0.04556591063737869,
+      "eval_runtime": 93.7142,
+      "eval_samples_per_second": 52.82,
+      "eval_steps_per_second": 0.416,
+      "eval_wer": 0.25333333333333335,
+      "step": 110000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.1001300276781274e-05,
+      "loss": 0.0258,
+      "step": 111000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.8967193188406938e-05,
+      "loss": 0.0255,
+      "step": 112000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.699495426678389e-05,
+      "loss": 0.0241,
+      "step": 113000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.5085591658461056e-05,
+      "loss": 0.0243,
+      "step": 114000
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.3240081369591984e-05,
+      "loss": 0.0249,
+      "step": 115000
+    },
+    {
+      "epoch": 8.22,
+      "eval_cer": 0.09116547517544435,
+      "eval_loss": 0.04581727460026741,
+      "eval_runtime": 93.453,
+      "eval_samples_per_second": 52.968,
+      "eval_steps_per_second": 0.417,
+      "eval_wer": 0.25292929292929295,
+      "step": 115000
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.1459366767031522e-05,
+      "loss": 0.0247,
+      "step": 116000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.9744358096116225e-05,
+      "loss": 0.0249,
+      "step": 117000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.8095932015375496e-05,
+      "loss": 0.0247,
+      "step": 118000
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.65149311484114e-05,
+      "loss": 0.0246,
+      "step": 119000
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.500216365317587e-05,
+      "loss": 0.025,
+      "step": 120000
+    },
+    {
+      "epoch": 8.57,
+      "eval_cer": 0.09134037296954592,
+      "eval_loss": 0.045206133276224136,
+      "eval_runtime": 93.5201,
+      "eval_samples_per_second": 52.93,
+      "eval_steps_per_second": 0.417,
+      "eval_wer": 0.25191919191919193,
+      "step": 120000
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.355840280886582e-05,
+      "loss": 0.025,
+      "step": 121000
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.2184386620647097e-05,
+      "loss": 0.0247,
+      "step": 122000
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.0880817442409478e-05,
+      "loss": 0.0248,
+      "step": 123000
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 9.648361617745371e-06,
+      "loss": 0.0248,
+      "step": 124000
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 8.487649139335962e-06,
+      "loss": 0.0248,
+      "step": 125000
+    },
+    {
+      "epoch": 8.93,
+      "eval_cer": 0.09169016855774907,
+      "eval_loss": 0.04524415731430054,
+      "eval_runtime": 93.2177,
+      "eval_samples_per_second": 53.101,
+      "eval_steps_per_second": 0.418,
+      "eval_wer": 0.25292929292929295,
+      "step": 125000
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.399273326918692e-06,
+      "loss": 0.0246,
+      "step": 126000
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 6.383790524001009e-06,
+      "loss": 0.024,
+      "step": 127000
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.441719813474849e-06,
+      "loss": 0.0241,
+      "step": 128000
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.57354275227797e-06,
+      "loss": 0.0242,
+      "step": 129000
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.7797031252369767e-06,
+      "loss": 0.0243,
+      "step": 130000
+    },
+    {
+      "epoch": 9.29,
+      "eval_cer": 0.09138409741807131,
+      "eval_loss": 0.0453341044485569,
+      "eval_runtime": 93.6155,
+      "eval_samples_per_second": 52.876,
+      "eval_steps_per_second": 0.417,
+      "eval_wer": 0.25212121212121213,
+      "step": 130000
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.0606067182186776e-06,
+      "loss": 0.024,
+      "step": 131000
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.4166211107049584e-06,
+      "loss": 0.0242,
+      "step": 132000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8480754878977489e-06,
+      "loss": 0.0241,
+      "step": 133000
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3552604724498928e-06,
+      "loss": 0.0241,
+      "step": 134000
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 9.384279759080127e-07,
+      "loss": 0.024,
+      "step": 135000
+    },
+    {
+      "epoch": 9.65,
+      "eval_cer": 0.09129664852102053,
+      "eval_loss": 0.04531262069940567,
+      "eval_runtime": 93.4097,
+      "eval_samples_per_second": 52.992,
+      "eval_steps_per_second": 0.418,
+      "eval_wer": 0.25212121212121213,
+      "step": 135000
+    }
+  ],
+  "max_steps": 139950,
+  "num_train_epochs": 10,
+  "total_flos": 4.19385405764928e+18,
+  "trial_name": null,
+  "trial_params": null
+}

byT5/Charsiu/g2p_multilingual_byT5_small_100/.gitattributes

@@ -0,0 +1,31 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

byT5/Charsiu/g2p_multilingual_byT5_small_100/charsiu_g2p_multilingual_byT5_small_100.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:082b8bcbe414728e20f46bf50e22e15db2c8b3487c96cf4afd9fbefe5d5f1d7f
+size 1198914720

byT5/Charsiu/g2p_multilingual_byT5_small_100/config.json

@@ -0,0 +1,32 @@
+{
+  "_name_or_path": "google/byt5-small",
+  "architectures": [
+    "T5ForConditionalGeneration"
+  ],
+  "d_ff": 3584,
+  "d_kv": 64,
+  "d_model": 1472,
+  "decoder_start_token_id": 0,
+  "dense_act_fn": "gelu_new",
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-gelu",
+  "gradient_checkpointing": false,
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "is_gated_act": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "num_decoder_layers": 4,
+  "num_heads": 6,
+  "num_layers": 12,
+  "pad_token_id": 0,
+  "relative_attention_max_distance": 128,
+  "relative_attention_num_buckets": 32,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "ByT5Tokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.21.1",
+  "use_cache": true,
+  "vocab_size": 384
+}

byT5/Charsiu/g2p_multilingual_byT5_small_100/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ef4032409d33944e4249f959d028f78436f968e26b81cb6b0d508a6cf12ce6c
+size 1198608045

byT5/Charsiu/g2p_multilingual_byT5_small_100/source.txt

@@ -0,0 +1,2 @@
+https://huggingface.co/charsiu/g2p_multilingual_byT5_small_100
+https://huggingface.co/Jarbas/charsiu_g2p_multilingual_byT5_small_100_onnx

byT5/Charsiu/g2p_multilingual_byT5_small_100/trainer_state.json

@@ -0,0 +1,1136 @@
+{
+  "best_metric": 0.04629843682050705,
+  "best_model_checkpoint": "/scratch/lingjzhu_root/lingjzhu1/lingjzhu/g2p/byt5_small_finetuned_from_pretrained/checkpoint-125000",
+  "epoch": 9.917799364213547,
+  "global_step": 140000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003,
+      "loss": 1.3905,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002999623214094654,
+      "loss": 0.2662,
+      "step": 2000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002998493045668776,
+      "loss": 0.185,
+      "step": 3000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029966100624977436,
+      "loss": 0.1523,
+      "step": 4000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029939752105569156,
+      "loss": 0.1309,
+      "step": 5000
+    },
+    {
+      "epoch": 0.35,
+      "eval_cer": 0.15177127579292352,
+      "eval_loss": 0.15828794240951538,
+      "eval_runtime": 99.5729,
+      "eval_samples_per_second": 50.214,
+      "eval_steps_per_second": 0.402,
+      "eval_wer": 0.4034,
+      "step": 5000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029905898135463893,
+      "loss": 0.1159,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002986455572226002,
+      "loss": 0.1057,
+      "step": 7000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029815745635609,
+      "loss": 0.0972,
+      "step": 8000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029759492396781067,
+      "loss": 0.0899,
+      "step": 9000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00029695824266346256,
+      "loss": 0.0849,
+      "step": 10000
+    },
+    {
+      "epoch": 0.71,
+      "eval_cer": 0.1211970935121499,
+      "eval_loss": 0.10069698095321655,
+      "eval_runtime": 99.7318,
+      "eval_samples_per_second": 50.134,
+      "eval_steps_per_second": 0.401,
+      "eval_wer": 0.3372,
+      "step": 10000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000296247732299768,
+      "loss": 0.0804,
+      "step": 11000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000295463749823781,
+      "loss": 0.0766,
+      "step": 12000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002946066890935644,
+      "loss": 0.0733,
+      "step": 13000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002936769806803222,
+      "loss": 0.0707,
+      "step": 14000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029267509165208945,
+      "loss": 0.0661,
+      "step": 15000
+    },
+    {
+      "epoch": 1.06,
+      "eval_cer": 0.11274499234567369,
+      "eval_loss": 0.08334551751613617,
+      "eval_runtime": 99.5349,
+      "eval_samples_per_second": 50.234,
+      "eval_steps_per_second": 0.402,
+      "eval_wer": 0.3146,
+      "step": 15000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00029160152533908566,
+      "loss": 0.0642,
+      "step": 16000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002904568210808509,
+      "loss": 0.0624,
+      "step": 17000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028924155395529237,
+      "loss": 0.061,
+      "step": 18000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028795633448977547,
+      "loss": 0.0593,
+      "step": 19000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002866018083544069,
+      "loss": 0.0585,
+      "step": 20000
+    },
+    {
+      "epoch": 1.42,
+      "eval_cer": 0.10940296254770478,
+      "eval_loss": 0.07396863400936127,
+      "eval_runtime": 99.5625,
+      "eval_samples_per_second": 50.22,
+      "eval_steps_per_second": 0.402,
+      "eval_wer": 0.3024,
+      "step": 20000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002851786560376614,
+      "loss": 0.0568,
+      "step": 21000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00028368759250451786,
+      "loss": 0.0552,
+      "step": 22000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00028212936683727404,
+      "loss": 0.0542,
+      "step": 23000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002805047618592216,
+      "loss": 0.0532,
+      "step": 24000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002788145937413704,
+      "loss": 0.0526,
+      "step": 25000
+    },
+    {
+      "epoch": 1.77,
+      "eval_cer": 0.10444382155717027,
+      "eval_loss": 0.06628572195768356,
+      "eval_runtime": 99.5351,
+      "eval_samples_per_second": 50.234,
+      "eval_steps_per_second": 0.402,
+      "eval_wer": 0.293,
+      "step": 25000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00027705971159241994,
+      "loss": 0.0515,
+      "step": 26000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002752409970321824,
+      "loss": 0.0505,
+      "step": 27000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000273359363748674,
+      "loss": 0.0501,
+      "step": 28000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002714157570390952,
+      "loss": 0.0471,
+      "step": 29000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002694111533349308,
+      "loss": 0.0459,
+      "step": 30000
+    },
+    {
+      "epoch": 2.13,
+      "eval_cer": 0.10159770585825481,
+      "eval_loss": 0.06294897198677063,
+      "eval_runtime": 99.7848,
+      "eval_samples_per_second": 50.108,
+      "eval_steps_per_second": 0.401,
+      "eval_wer": 0.2848,
+      "step": 30000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00026734655971140956,
+      "loss": 0.0455,
+      "step": 31000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026522301338156835,
+      "loss": 0.045,
+      "step": 32000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00026304158117517563,
+      "loss": 0.0448,
+      "step": 33000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002608033590027766,
+      "loss": 0.0441,
+      "step": 34000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000258509471305128,
+      "loss": 0.0436,
+      "step": 35000
+    },
+    {
+      "epoch": 2.48,
+      "eval_cer": 0.10114491472433644,
+      "eval_loss": 0.06108950078487396,
+      "eval_runtime": 99.5917,
+      "eval_samples_per_second": 50.205,
+      "eval_steps_per_second": 0.402,
+      "eval_wer": 0.2796,
+      "step": 35000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00025616107048830045,
+      "loss": 0.0431,
+      "step": 36000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002537593363447311,
+      "loss": 0.043,
+      "step": 37000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002513054754605181,
+      "loss": 0.0427,
+      "step": 38000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002488007206092547,
+      "loss": 0.0418,
+      "step": 39000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00024624633013270664,
+      "loss": 0.0417,
+      "step": 40000
+    },
+    {
+      "epoch": 2.83,
+      "eval_cer": 0.09972185687487872,
+      "eval_loss": 0.05735113471746445,
+      "eval_runtime": 99.9084,
+      "eval_samples_per_second": 50.046,
+      "eval_steps_per_second": 0.4,
+      "eval_wer": 0.2768,
+      "step": 40000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002436435873086449,
+      "loss": 0.0411,
+      "step": 41000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00024099379970615137,
+      "loss": 0.0406,
+      "step": 42000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00023829829852872037,
+      "loss": 0.0386,
+      "step": 43000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00023555843794548724,
+      "loss": 0.0375,
+      "step": 44000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00023277559441091917,
+      "loss": 0.0376,
+      "step": 45000
+    },
+    {
+      "epoch": 3.19,
+      "eval_cer": 0.09982966428771642,
+      "eval_loss": 0.05706231668591499,
+      "eval_runtime": 100.0238,
+      "eval_samples_per_second": 49.988,
+      "eval_steps_per_second": 0.4,
+      "eval_wer": 0.2766,
+      "step": 45000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002299511659733103,
+      "loss": 0.0373,
+      "step": 46000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002270865715724285,
+      "loss": 0.0371,
+      "step": 47000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0002241832503266669,
+      "loss": 0.0368,
+      "step": 48000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00022124266081005755,
+      "loss": 0.0366,
+      "step": 49000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0002182662803195115,
+      "loss": 0.0364,
+      "step": 50000
+    },
+    {
+      "epoch": 3.54,
+      "eval_cer": 0.09832036050798854,
+      "eval_loss": 0.05559751018881798,
+      "eval_runtime": 100.1039,
+      "eval_samples_per_second": 49.948,
+      "eval_steps_per_second": 0.4,
+      "eval_wer": 0.2714,
+      "step": 50000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00021525560413265246,
+      "loss": 0.0363,
+      "step": 51000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00021221214475661745,
+      "loss": 0.0358,
+      "step": 52000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020913743116820165,
+      "loss": 0.0358,
+      "step": 53000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020603300804572917,
+      "loss": 0.0356,
+      "step": 54000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00020290043499303573,
+      "loss": 0.0353,
+      "step": 55000
+    },
+    {
+      "epoch": 3.9,
+      "eval_cer": 0.09775976196123246,
+      "eval_loss": 0.053284887224435806,
+      "eval_runtime": 99.9875,
+      "eval_samples_per_second": 50.006,
+      "eval_steps_per_second": 0.4,
+      "eval_wer": 0.2718,
+      "step": 55000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019974128575595291,
+      "loss": 0.0349,
+      "step": 56000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.000196557147431688,
+      "loss": 0.0335,
+      "step": 57000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019334961967149597,
+      "loss": 0.0324,
+      "step": 58000
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00019012031387704483,
+      "loss": 0.0323,
+      "step": 59000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018687085239087786,
+      "loss": 0.0323,
+      "step": 60000
+    },
+    {
+      "epoch": 4.25,
+      "eval_cer": 0.0979969382694754,
+      "eval_loss": 0.05237960070371628,
+      "eval_runtime": 99.6271,
+      "eval_samples_per_second": 50.187,
+      "eval_steps_per_second": 0.401,
+      "eval_wer": 0.2696,
+      "step": 60000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018360286768137894,
+      "loss": 0.0322,
+      "step": 61000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018031800152265137,
+      "loss": 0.032,
+      "step": 62000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00017701790416972107,
+      "loss": 0.0319,
+      "step": 63000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017370423352948002,
+      "loss": 0.0318,
+      "step": 64000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017037865432778442,
+      "loss": 0.0316,
+      "step": 65000
+    },
+    {
+      "epoch": 4.6,
+      "eval_cer": 0.09584079001272128,
+      "eval_loss": 0.052747271955013275,
+      "eval_runtime": 99.8825,
+      "eval_samples_per_second": 50.059,
+      "eval_steps_per_second": 0.4,
+      "eval_wer": 0.2678,
+      "step": 65000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00016704283727312792,
+      "loss": 0.0315,
+      "step": 66000
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00016369845821730933,
+      "loss": 0.0314,
+      "step": 67000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016034719731351576,
+      "loss": 0.0314,
+      "step": 68000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00015699073817224542,
+      "loss": 0.0312,
+      "step": 69000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00015363076701549341,
+      "loss": 0.031,
+      "step": 70000
+    },
+    {
+      "epoch": 4.96,
+      "eval_cer": 0.09558205222191078,
+      "eval_loss": 0.05028079077601433,
+      "eval_runtime": 99.8117,
+      "eval_samples_per_second": 50.094,
+      "eval_steps_per_second": 0.401,
+      "eval_wer": 0.2664,
+      "step": 70000
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00015026897182962535,
+      "loss": 0.0303,
+      "step": 71000
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00014690704151736484,
+      "loss": 0.0286,
+      "step": 72000
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0001435466650493204,
+      "loss": 0.0287,
+      "step": 73000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0001401895306154785,
+      "loss": 0.0284,
+      "step": 74000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0001368373247770885,
+      "loss": 0.0287,
+      "step": 75000
+    },
+    {
+      "epoch": 5.31,
+      "eval_cer": 0.09566829815218095,
+      "eval_loss": 0.04991844296455383,
+      "eval_runtime": 99.6863,
+      "eval_samples_per_second": 50.157,
+      "eval_steps_per_second": 0.401,
+      "eval_wer": 0.2642,
+      "step": 75000
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00013349173161936601,
+      "loss": 0.0287,
+      "step": 76000
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00013015443190544018,
+      "loss": 0.0283,
+      "step": 77000
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00012682710223196978,
+      "loss": 0.0284,
+      "step": 78000
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00012351141418685273,
+      "loss": 0.0286,
+      "step": 79000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00012020903350945151,
+      "loss": 0.0282,
+      "step": 80000
+    },
+    {
+      "epoch": 5.67,
+      "eval_cer": 0.09588391297785635,
+      "eval_loss": 0.04885885864496231,
+      "eval_runtime": 99.7588,
+      "eval_samples_per_second": 50.121,
+      "eval_steps_per_second": 0.401,
+      "eval_wer": 0.264,
+      "step": 80000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00011692161925375738,
+      "loss": 0.0282,
+      "step": 81000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00011365082295491233,
+      "loss": 0.0279,
+      "step": 82000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00011039828779950928,
+      "loss": 0.0281,
+      "step": 83000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0001071656478000858,
+      "loss": 0.028,
+      "step": 84000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00010395452697422703,
+      "loss": 0.0275,
+      "step": 85000
+    },
+    {
+      "epoch": 6.02,
+      "eval_cer": 0.0948058388494793,
+      "eval_loss": 0.048800304532051086,
+      "eval_runtime": 99.8744,
+      "eval_samples_per_second": 50.063,
+      "eval_steps_per_second": 0.401,
+      "eval_wer": 0.2616,
+      "step": 85000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00010076653852868962,
+      "loss": 0.026,
+      "step": 86000
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 9.760328404895728e-05,
+      "loss": 0.0258,
+      "step": 87000
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 9.446635269463389e-05,
+      "loss": 0.0258,
+      "step": 88000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 9.135732040107998e-05,
+      "loss": 0.026,
+      "step": 89000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 8.82777490876921e-05,
+      "loss": 0.0259,
+      "step": 90000
+    },
+    {
+      "epoch": 6.38,
+      "eval_cer": 0.09374932620366977,
+      "eval_loss": 0.0478358194231987,
+      "eval_runtime": 99.8791,
+      "eval_samples_per_second": 50.061,
+      "eval_steps_per_second": 0.4,
+      "eval_wer": 0.2608,
+      "step": 90000
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 8.522918587322418e-05,
+      "loss": 0.0259,
+      "step": 91000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 8.22131622985438e-05,
+      "loss": 0.026,
+      "step": 92000
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.923119355721526e-05,
+      "loss": 0.0259,
+      "step": 93000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.628477773429438e-05,
+      "loss": 0.0257,
+      "step": 94000
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.3375395053719e-05,
+      "loss": 0.0257,
+      "step": 95000
+    },
+    {
+      "epoch": 6.73,
+      "eval_cer": 0.09379244916880485,
+      "eval_loss": 0.04708043858408928,
+      "eval_runtime": 99.9161,
+      "eval_samples_per_second": 50.042,
+      "eval_steps_per_second": 0.4,
+      "eval_wer": 0.259,
+      "step": 95000
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 7.050450713467216e-05,
+      "loss": 0.0255,
+      "step": 96000
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.767355625729215e-05,
+      "loss": 0.0255,
+      "step": 97000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.488396463809805e-05,
+      "loss": 0.0255,
+      "step": 98000
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.213713371549497e-05,
+      "loss": 0.025,
+      "step": 99000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.943444344571755e-05,
+      "loss": 0.0238,
+      "step": 100000
+    },
+    {
+      "epoch": 7.08,
+      "eval_cer": 0.0940296254770478,
+      "eval_loss": 0.047569289803504944,
+      "eval_runtime": 99.6286,
+      "eval_samples_per_second": 50.186,
+      "eval_steps_per_second": 0.401,
+      "eval_wer": 0.2586,
+      "step": 100000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.6777251609566e-05,
+      "loss": 0.0242,
+      "step": 101000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.416689313028253e-05,
+      "loss": 0.0239,
+      "step": 102000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.1604679402911015e-05,
+      "loss": 0.024,
+      "step": 103000
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.9091897635476726e-05,
+      "loss": 0.0242,
+      "step": 104000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.6629810202317166e-05,
+      "loss": 0.0239,
+      "step": 105000
+    },
+    {
+      "epoch": 7.44,
+      "eval_cer": 0.09372776472110222,
+      "eval_loss": 0.04707524552941322,
+      "eval_runtime": 99.582,
+      "eval_samples_per_second": 50.21,
+      "eval_steps_per_second": 0.402,
+      "eval_wer": 0.2576,
+      "step": 105000
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.421965400988918e-05,
+      "loss": 0.0239,
+      "step": 106000
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.1862639875370016e-05,
+      "loss": 0.024,
+      "step": 107000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.955995191836581e-05,
+      "loss": 0.0238,
+      "step": 108000
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.731274696603201e-05,
+      "loss": 0.0239,
+      "step": 109000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.512215397190527e-05,
+      "loss": 0.0237,
+      "step": 110000
+    },
+    {
+      "epoch": 7.79,
+      "eval_cer": 0.09361995730826451,
+      "eval_loss": 0.046346429735422134,
+      "eval_runtime": 99.9122,
+      "eval_samples_per_second": 50.044,
+      "eval_steps_per_second": 0.4,
+      "eval_wer": 0.2564,
+      "step": 110000
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.298927344873821e-05,
+      "loss": 0.0237,
+      "step": 111000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.091517691562284e-05,
+      "loss": 0.0237,
+      "step": 112000
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.890090635967912e-05,
+      "loss": 0.024,
+      "step": 113000
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.694747371258042e-05,
+      "loss": 0.0227,
+      "step": 114000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.505586034217802e-05,
+      "loss": 0.0226,
+      "step": 115000
+    },
+    {
+      "epoch": 8.15,
+      "eval_cer": 0.0940296254770478,
+      "eval_loss": 0.0467340350151062,
+      "eval_runtime": 100.0479,
+      "eval_samples_per_second": 49.976,
+      "eval_steps_per_second": 0.4,
+      "eval_wer": 0.2582,
+      "step": 115000
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.3227016559480356e-05,
+      "loss": 0.0228,
+      "step": 116000
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.1461861141234628e-05,
+      "loss": 0.0226,
+      "step": 117000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.9761280868350654e-05,
+      "loss": 0.0228,
+      "step": 118000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.8126130080398725e-05,
+      "loss": 0.0226,
+      "step": 119000
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.655723024640553e-05,
+      "loss": 0.0228,
+      "step": 120000
+    },
+    {
+      "epoch": 8.5,
+      "eval_cer": 0.09323185062204878,
+      "eval_loss": 0.04636606574058533,
+      "eval_runtime": 99.8629,
+      "eval_samples_per_second": 50.069,
+      "eval_steps_per_second": 0.401,
+      "eval_wer": 0.255,
+      "step": 120000
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.5055369552163509e-05,
+      "loss": 0.0228,
+      "step": 121000
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.3621302504261194e-05,
+      "loss": 0.0229,
+      "step": 122000
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.2255749551033205e-05,
+      "loss": 0.0227,
+      "step": 123000
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.0959396720620555e-05,
+      "loss": 0.0226,
+      "step": 124000
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 9.732895276323099e-06,
+      "loss": 0.0227,
+      "step": 125000
+    },
+    {
+      "epoch": 8.86,
+      "eval_cer": 0.0935552728605619,
+      "eval_loss": 0.04629843682050705,
+      "eval_runtime": 100.0662,
+      "eval_samples_per_second": 49.967,
+      "eval_steps_per_second": 0.4,
+      "eval_wer": 0.2578,
+      "step": 125000
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 8.576861389416939e-06,
+      "loss": 0.0227,
+      "step": 126000
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.4918758296016734e-06,
+      "loss": 0.0227,
+      "step": 127000
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 6.478483673232537e-06,
+      "loss": 0.0223,
+      "step": 128000
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.537194029484349e-06,
+      "loss": 0.0222,
+      "step": 129000
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.668479784584583e-06,
+      "loss": 0.0221,
+      "step": 130000
+    },
+    {
+      "epoch": 9.21,
+      "eval_cer": 0.09361995730826451,
+      "eval_loss": 0.04640863090753555,
+      "eval_runtime": 99.7393,
+      "eval_samples_per_second": 50.131,
+      "eval_steps_per_second": 0.401,
+      "eval_wer": 0.2562,
+      "step": 130000
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.872777364244245e-06,
+      "loss": 0.0223,
+      "step": 131000
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.150486514405798e-06,
+      "loss": 0.0222,
+      "step": 132000
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.50197010041826e-06,
+      "loss": 0.022,
+      "step": 133000
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.9275539247405437e-06,
+      "loss": 0.0222,
+      "step": 134000
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4275265632643806e-06,
+      "loss": 0.0223,
+      "step": 135000
+    },
+    {
+      "epoch": 9.56,
+      "eval_cer": 0.09357683434312944,
+      "eval_loss": 0.046345390379428864,
+      "eval_runtime": 99.9431,
+      "eval_samples_per_second": 50.028,
+      "eval_steps_per_second": 0.4,
+      "eval_wer": 0.2564,
+      "step": 135000
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.0021392203391932e-06,
+      "loss": 0.022,
+      "step": 136000
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 6.516056025718585e-07,
+      "loss": 0.0221,
+      "step": 137000
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.76101811464391e-07,
+      "loss": 0.0223,
+      "step": 138000
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.7576625494392648e-07,
+      "loss": 0.0221,
+      "step": 139000
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.0699577829210744e-08,
+      "loss": 0.0223,
+      "step": 140000
+    },
+    {
+      "epoch": 9.92,
+      "eval_cer": 0.0937708876862373,
+      "eval_loss": 0.046362996101379395,
+      "eval_runtime": 99.8262,
+      "eval_samples_per_second": 50.087,
+      "eval_steps_per_second": 0.401,
+      "eval_wer": 0.2566,
+      "step": 140000
+    }
+  ],
+  "max_steps": 141160,
+  "num_train_epochs": 10,
+  "total_flos": 4.3451206845736243e+18,
+  "trial_name": null,
+  "trial_params": null
+}

byT5/Charsiu/g2p_multilingual_byT5_tiny_12_layers/.gitattributes

@@ -0,0 +1,27 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

byT5/Charsiu/g2p_multilingual_byT5_tiny_12_layers/config.json

@@ -0,0 +1,29 @@
+{
+  "_name_or_path": "/home/patrick/t5/byt5-small",
+  "architectures": [
+    "T5ForConditionalGeneration"
+  ],
+  "d_ff": 1024,
+  "d_kv": 64,
+  "d_model": 256,
+  "decoder_start_token_id": 0,
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-gelu",
+  "gradient_checkpointing": false,
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "num_decoder_layers": 4,
+  "num_heads": 6,
+  "num_layers": 8,
+  "pad_token_id": 0,
+  "relative_attention_num_buckets": 32,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "ByT5Tokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.16.2",
+  "use_cache": true,
+  "vocab_size": 384
+}

byT5/Charsiu/g2p_multilingual_byT5_tiny_12_layers/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf8b9f8b869324c3377a9a61a1d7235f9361ef212cf96b00f8533f417564c84f
+size 63793205

byT5/Charsiu/g2p_multilingual_byT5_tiny_12_layers/source.txt

@@ -0,0 +1 @@
+https://huggingface.co/charsiu/g2p_multilingual_byT5_tiny_12_layers

byT5/Charsiu/g2p_multilingual_byT5_tiny_12_layers/trainer_state.json

@@ -0,0 +1,1096 @@
+{
+  "best_metric": 0.2115170197415885,
+  "best_model_checkpoint": "/scratch/lingjzhu_root/lingjzhu1/lingjzhu/g2p/byt5_12_layers_baseline/checkpoint-5000",
+  "epoch": 9.646302250803858,
+  "global_step": 135000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003,
+      "loss": 2.113,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002999616623572683,
+      "loss": 0.5915,
+      "step": 2000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029984666902607135,
+      "loss": 0.3074,
+      "step": 3000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002996550787873857,
+      "loss": 0.2407,
+      "step": 4000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002993869895761197,
+      "loss": 0.2091,
+      "step": 5000
+    },
+    {
+      "epoch": 0.36,
+      "eval_cer": 0.2115170197415885,
+      "eval_loss": 0.2637424170970917,
+      "eval_runtime": 81.6578,
+      "eval_samples_per_second": 60.619,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.5187878787878788,
+      "step": 5000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000299042538431052,
+      "loss": 0.1885,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029862190142478177,
+      "loss": 0.1748,
+      "step": 7000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029812529357372587,
+      "loss": 0.1637,
+      "step": 8000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029755296872820933,
+      "loss": 0.1547,
+      "step": 9000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002969052194427048,
+      "loss": 0.1471,
+      "step": 10000
+    },
+    {
+      "epoch": 0.71,
+      "eval_cer": 0.16525655320172275,
+      "eval_loss": 0.1778797060251236,
+      "eval_runtime": 81.5727,
+      "eval_samples_per_second": 60.682,
+      "eval_steps_per_second": 0.123,
+      "eval_wer": 0.4212121212121212,
+      "step": 10000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002961823768262882,
+      "loss": 0.1408,
+      "step": 11000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002953848103733858,
+      "loss": 0.1341,
+      "step": 12000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029451292777490066,
+      "loss": 0.1294,
+      "step": 13000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002935671747098137,
+      "loss": 0.1233,
+      "step": 14000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029254803461736643,
+      "loss": 0.1179,
+      "step": 15000
+    },
+    {
+      "epoch": 1.07,
+      "eval_cer": 0.1474607026518878,
+      "eval_loss": 0.14176948368549347,
+      "eval_runtime": 81.7933,
+      "eval_samples_per_second": 60.518,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.3838383838383838,
+      "step": 15000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00029145602844994243,
+      "loss": 0.1143,
+      "step": 16000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002902917144067724,
+      "loss": 0.1107,
+      "step": 17000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028905568764860047,
+      "loss": 0.107,
+      "step": 18000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028774857999345685,
+      "loss": 0.1051,
+      "step": 19000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002863710595936922,
+      "loss": 0.1016,
+      "step": 20000
+    },
+    {
+      "epoch": 1.43,
+      "eval_cer": 0.1344308169913207,
+      "eval_loss": 0.1187734305858612,
+      "eval_runtime": 81.9499,
+      "eval_samples_per_second": 60.403,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.36404040404040405,
+      "step": 20000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002849238305944389,
+      "loss": 0.0988,
+      "step": 21000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00028340763277367477,
+      "loss": 0.0972,
+      "step": 22000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002818232411640713,
+      "loss": 0.0953,
+      "step": 23000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00028017146565682144,
+      "loss": 0.0933,
+      "step": 24000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00027845315058764886,
+      "loss": 0.0904,
+      "step": 25000
+    },
+    {
+      "epoch": 1.79,
+      "eval_cer": 0.12669158960232615,
+      "eval_loss": 0.10707778483629227,
+      "eval_runtime": 81.9487,
+      "eval_samples_per_second": 60.404,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.3466666666666667,
+      "step": 25000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00027666917430520975,
+      "loss": 0.089,
+      "step": 26000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00027482044872210895,
+      "loss": 0.0869,
+      "step": 27000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002729079188487587,
+      "loss": 0.0856,
+      "step": 28000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00027093256231031885,
+      "loss": 0.0827,
+      "step": 29000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00026889538884696597,
+      "loss": 0.082,
+      "step": 30000
+    },
+    {
+      "epoch": 2.14,
+      "eval_cer": 0.1201110600992545,
+      "eval_loss": 0.09880149364471436,
+      "eval_runtime": 81.9937,
+      "eval_samples_per_second": 60.371,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.3375757575757576,
+      "step": 30000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002667974397977457,
+      "loss": 0.0803,
+      "step": 31000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002646397875682729,
+      "loss": 0.0794,
+      "step": 32000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00026242353508255185,
+      "loss": 0.0783,
+      "step": 33000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002601498152191957,
+      "loss": 0.078,
+      "step": 34000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002578197902323352,
+      "loss": 0.0765,
+      "step": 35000
+    },
+    {
+      "epoch": 2.5,
+      "eval_cer": 0.11654751754443497,
+      "eval_loss": 0.09110759943723679,
+      "eval_runtime": 82.1822,
+      "eval_samples_per_second": 60.232,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.32545454545454544,
+      "step": 35000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00025543465115751026,
+      "loss": 0.0749,
+      "step": 36000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0002529956172028505,
+      "loss": 0.0743,
+      "step": 37000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002505039351258541,
+      "loss": 0.0732,
+      "step": 38000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002479608785960846,
+      "loss": 0.0722,
+      "step": 39000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0002453677475441111,
+      "loss": 0.0716,
+      "step": 40000
+    },
+    {
+      "epoch": 2.86,
+      "eval_cer": 0.11418639732406376,
+      "eval_loss": 0.0829671323299408,
+      "eval_runtime": 81.5052,
+      "eval_samples_per_second": 60.732,
+      "eval_steps_per_second": 0.123,
+      "eval_wer": 0.3208080808080808,
+      "step": 40000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00024272586749702474,
+      "loss": 0.0704,
+      "step": 41000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002400365889008706,
+      "loss": 0.0702,
+      "step": 42000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023730128643034235,
+      "loss": 0.0673,
+      "step": 43000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00023452135828609167,
+      "loss": 0.0676,
+      "step": 44000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0002316982254800121,
+      "loss": 0.0671,
+      "step": 45000
+    },
+    {
+      "epoch": 3.22,
+      "eval_cer": 0.11213134824337027,
+      "eval_loss": 0.08274847269058228,
+      "eval_runtime": 81.7806,
+      "eval_samples_per_second": 60.528,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.317979797979798,
+      "step": 45000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00022883333110886237,
+      "loss": 0.066,
+      "step": 46000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00022592813961660067,
+      "loss": 0.0656,
+      "step": 47000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00022298413604580696,
+      "loss": 0.0646,
+      "step": 48000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00022000282527857588,
+      "loss": 0.0643,
+      "step": 49000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0002169857312672683,
+      "loss": 0.0638,
+      "step": 50000
+    },
+    {
+      "epoch": 3.57,
+      "eval_cer": 0.11206576157058219,
+      "eval_loss": 0.07710296660661697,
+      "eval_runtime": 81.5659,
+      "eval_samples_per_second": 60.687,
+      "eval_steps_per_second": 0.123,
+      "eval_wer": 0.3113131313131313,
+      "step": 50000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00021393439625551483,
+      "loss": 0.0633,
+      "step": 51000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00021085037998986924,
+      "loss": 0.0627,
+      "step": 52000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020773525892251514,
+      "loss": 0.0624,
+      "step": 53000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00020459062540543316,
+      "loss": 0.0618,
+      "step": 54000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00020141808687644067,
+      "loss": 0.0613,
+      "step": 55000
+    },
+    {
+      "epoch": 3.93,
+      "eval_cer": 0.10915808574364355,
+      "eval_loss": 0.07446856051683426,
+      "eval_runtime": 81.6128,
+      "eval_samples_per_second": 60.652,
+      "eval_steps_per_second": 0.123,
+      "eval_wer": 0.30626262626262624,
+      "step": 55000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019821926503751995,
+      "loss": 0.0604,
+      "step": 56000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019499579502585537,
+      "loss": 0.059,
+      "step": 57000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019174932457800242,
+      "loss": 0.0584,
+      "step": 58000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001884815131876167,
+      "loss": 0.0581,
+      "step": 59000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018519403125717278,
+      "loss": 0.0574,
+      "step": 60000
+    },
+    {
+      "epoch": 4.29,
+      "eval_cer": 0.1064034454865438,
+      "eval_loss": 0.07090297341346741,
+      "eval_runtime": 81.8166,
+      "eval_samples_per_second": 60.501,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.2991919191919192,
+      "step": 60000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018188855924410722,
+      "loss": 0.0576,
+      "step": 61000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00017856678680182127,
+      "loss": 0.0569,
+      "step": 62000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001752304119159834,
+      "loss": 0.0568,
+      "step": 63000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017188114003657205,
+      "loss": 0.0566,
+      "step": 64000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00016852068320610358,
+      "loss": 0.0561,
+      "step": 65000
+    },
+    {
+      "epoch": 4.64,
+      "eval_cer": 0.10653461883211998,
+      "eval_loss": 0.06934704631567001,
+      "eval_runtime": 81.8952,
+      "eval_samples_per_second": 60.443,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.298989898989899,
+      "step": 65000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00016515075918448972,
+      "loss": 0.0555,
+      "step": 66000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016177309057097285,
+      "loss": 0.0548,
+      "step": 67000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00015838940392358722,
+      "loss": 0.0548,
+      "step": 68000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00015500142887659688,
+      "loss": 0.0543,
+      "step": 69000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00015161089725636095,
+      "loss": 0.0542,
+      "step": 70000
+    },
+    {
+      "epoch": 5.0,
+      "eval_cer": 0.1039767385933845,
+      "eval_loss": 0.0671503096818924,
+      "eval_runtime": 81.8677,
+      "eval_samples_per_second": 60.463,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.2913131313131313,
+      "step": 70000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00014821954219607845,
+      "loss": 0.0521,
+      "step": 71000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0001448290972498651,
+      "loss": 0.0522,
+      "step": 72000
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00014144129550661485,
+      "loss": 0.0519,
+      "step": 73000
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0001380578687040995,
+      "loss": 0.0516,
+      "step": 74000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00013468054634375843,
+      "loss": 0.0518,
+      "step": 75000
+    },
+    {
+      "epoch": 5.36,
+      "eval_cer": 0.1052228853763582,
+      "eval_loss": 0.06576833128929138,
+      "eval_runtime": 81.8774,
+      "eval_samples_per_second": 60.456,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.29555555555555557,
+      "step": 75000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00013131105480663235,
+      "loss": 0.0511,
+      "step": 76000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.000127951116470891,
+      "loss": 0.0512,
+      "step": 77000
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00012460244883140783,
+      "loss": 0.0506,
+      "step": 78000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0001212667636218309,
+      "loss": 0.0502,
+      "step": 79000
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00011794576593959775,
+      "loss": 0.05,
+      "step": 80000
+    },
+    {
+      "epoch": 5.72,
+      "eval_cer": 0.1032990096412409,
+      "eval_loss": 0.06467730551958084,
+      "eval_runtime": 81.823,
+      "eval_samples_per_second": 60.496,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.2901010101010101,
+      "step": 80000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00011464115337434394,
+      "loss": 0.0501,
+      "step": 81000
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00011135461514014796,
+      "loss": 0.0492,
+      "step": 82000
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00010808783121205837,
+      "loss": 0.0494,
+      "step": 83000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00010484247146734352,
+      "loss": 0.0491,
+      "step": 84000
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00010162019483190237,
+      "loss": 0.0472,
+      "step": 85000
+    },
+    {
+      "epoch": 6.07,
+      "eval_cer": 0.10340832076255438,
+      "eval_loss": 0.06243439018726349,
+      "eval_runtime": 82.0022,
+      "eval_samples_per_second": 60.364,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.2898989898989899,
+      "step": 85000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 9.842264843227404e-05,
+      "loss": 0.0476,
+      "step": 86000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 9.52514667536784e-05,
+      "loss": 0.0469,
+      "step": 87000
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 9.210827080451842e-05,
+      "loss": 0.047,
+      "step": 88000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 8.899466728777203e-05,
+      "loss": 0.0474,
+      "step": 89000
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 8.591224777969557e-05,
+      "loss": 0.0467,
+      "step": 90000
+    },
+    {
+      "epoch": 6.43,
+      "eval_cer": 0.10150630725169979,
+      "eval_loss": 0.061206888407468796,
+      "eval_runtime": 81.7254,
+      "eval_samples_per_second": 60.569,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.28929292929292927,
+      "step": 90000
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 8.286258791626041e-05,
+      "loss": 0.0465,
+      "step": 91000
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.984724658773716e-05,
+      "loss": 0.0464,
+      "step": 92000
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.686776514184009e-05,
+      "loss": 0.0462,
+      "step": 93000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.392566659583846e-05,
+      "loss": 0.046,
+      "step": 94000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 7.102245485803813e-05,
+      "loss": 0.0461,
+      "step": 95000
+    },
+    {
+      "epoch": 6.79,
+      "eval_cer": 0.10109092499070856,
+      "eval_loss": 0.06065311282873154,
+      "eval_runtime": 82.0331,
+      "eval_samples_per_second": 60.342,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.28383838383838383,
+      "step": 95000
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.81596139590308e-05,
+      "loss": 0.0457,
+      "step": 96000
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.533860729310434e-05,
+      "loss": 0.0455,
+      "step": 97000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.256087687020127e-05,
+      "loss": 0.0454,
+      "step": 98000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.98278425788092e-05,
+      "loss": 0.044,
+      "step": 99000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.71409014601578e-05,
+      "loss": 0.0439,
+      "step": 100000
+    },
+    {
+      "epoch": 7.15,
+      "eval_cer": 0.10098161386939507,
+      "eval_loss": 0.05971471220254898,
+      "eval_runtime": 82.6482,
+      "eval_samples_per_second": 59.892,
+      "eval_steps_per_second": 0.121,
+      "eval_wer": 0.2804040404040404,
+      "step": 100000
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.4501426994095876e-05,
+      "loss": 0.0436,
+      "step": 101000
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.191076839701103e-05,
+      "loss": 0.0436,
+      "step": 102000
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.9370249932153075e-05,
+      "loss": 0.0438,
+      "step": 103000
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.6881170232712164e-05,
+      "loss": 0.0437,
+      "step": 104000
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.444480163799822e-05,
+      "loss": 0.0435,
+      "step": 105000
+    },
+    {
+      "epoch": 7.5,
+      "eval_cer": 0.10120023611202204,
+      "eval_loss": 0.058374855667352676,
+      "eval_runtime": 81.7698,
+      "eval_samples_per_second": 60.536,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.28383838383838383,
+      "step": 105000
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.2062389543061265e-05,
+      "loss": 0.0433,
+      "step": 106000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.9735151762084384e-05,
+      "loss": 0.0434,
+      "step": 107000
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.746427790587557e-05,
+      "loss": 0.0428,
+      "step": 108000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.525092877377602e-05,
+      "loss": 0.0432,
+      "step": 109000
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.309623576029597e-05,
+      "loss": 0.0427,
+      "step": 110000
+    },
+    {
+      "epoch": 7.86,
+      "eval_cer": 0.10012898712314991,
+      "eval_loss": 0.05760599672794342,
+      "eval_runtime": 81.886,
+      "eval_samples_per_second": 60.45,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.2804040404040404,
+      "step": 110000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.1001300276781274e-05,
+      "loss": 0.043,
+      "step": 111000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.8967193188406938e-05,
+      "loss": 0.0424,
+      "step": 112000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.699495426678389e-05,
+      "loss": 0.042,
+      "step": 113000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.5085591658461056e-05,
+      "loss": 0.0416,
+      "step": 114000
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.3240081369591984e-05,
+      "loss": 0.0415,
+      "step": 115000
+    },
+    {
+      "epoch": 8.22,
+      "eval_cer": 0.10036947159003957,
+      "eval_loss": 0.05732354149222374,
+      "eval_runtime": 81.9536,
+      "eval_samples_per_second": 60.4,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.2797979797979798,
+      "step": 115000
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.1459366767031522e-05,
+      "loss": 0.0417,
+      "step": 116000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.9744358096116225e-05,
+      "loss": 0.0414,
+      "step": 117000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.8095932015375496e-05,
+      "loss": 0.0412,
+      "step": 118000
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.65149311484114e-05,
+      "loss": 0.0412,
+      "step": 119000
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.500216365317587e-05,
+      "loss": 0.0412,
+      "step": 120000
+    },
+    {
+      "epoch": 8.57,
+      "eval_cer": 0.09988850265626024,
+      "eval_loss": 0.05689243599772453,
+      "eval_runtime": 81.8151,
+      "eval_samples_per_second": 60.502,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.2783838383838384,
+      "step": 120000
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.355840280886582e-05,
+      "loss": 0.0414,
+      "step": 121000
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.2184386620647097e-05,
+      "loss": 0.0416,
+      "step": 122000
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.0880817442409478e-05,
+      "loss": 0.0413,
+      "step": 123000
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 9.648361617745371e-06,
+      "loss": 0.0409,
+      "step": 124000
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 8.487649139335962e-06,
+      "loss": 0.0412,
+      "step": 125000
+    },
+    {
+      "epoch": 8.93,
+      "eval_cer": 0.09986664043199756,
+      "eval_loss": 0.05683436617255211,
+      "eval_runtime": 82.2238,
+      "eval_samples_per_second": 60.202,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.2781818181818182,
+      "step": 125000
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.399273326918692e-06,
+      "loss": 0.0411,
+      "step": 126000
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 6.383790524001009e-06,
+      "loss": 0.0407,
+      "step": 127000
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.441719813474849e-06,
+      "loss": 0.0406,
+      "step": 128000
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.57354275227797e-06,
+      "loss": 0.0407,
+      "step": 129000
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.7797031252369767e-06,
+      "loss": 0.0401,
+      "step": 130000
+    },
+    {
+      "epoch": 9.29,
+      "eval_cer": 0.09988850265626024,
+      "eval_loss": 0.05650737136602402,
+      "eval_runtime": 81.8641,
+      "eval_samples_per_second": 60.466,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.2777777777777778,
+      "step": 130000
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.0606067182186776e-06,
+      "loss": 0.0403,
+      "step": 131000
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.4166211107049584e-06,
+      "loss": 0.0406,
+      "step": 132000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8480754878977489e-06,
+      "loss": 0.0406,
+      "step": 133000
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3552604724498928e-06,
+      "loss": 0.0406,
+      "step": 134000
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 9.384279759080127e-07,
+      "loss": 0.0405,
+      "step": 135000
+    },
+    {
+      "epoch": 9.65,
+      "eval_cer": 0.09984477820773485,
+      "eval_loss": 0.056535448879003525,
+      "eval_runtime": 81.6573,
+      "eval_samples_per_second": 60.619,
+      "eval_steps_per_second": 0.122,
+      "eval_wer": 0.2781818181818182,
+      "step": 135000
+    }
+  ],
+  "max_steps": 139950,
+  "num_train_epochs": 10,
+  "total_flos": 2.638176399703204e+17,
+  "trial_name": null,
+  "trial_params": null
+}

byT5/Charsiu/g2p_multilingual_byT5_tiny_12_layers_100/.gitattributes

@@ -0,0 +1,31 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

byT5/Charsiu/g2p_multilingual_byT5_tiny_12_layers_100/charsiu_g2p_multilingual_byT5_tiny_12_layers_100.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:212cebd08d94633bf0839bc5ea5b137e5a3374df3d5fea08214e9a37ce25426e
+size 64030145

byT5/Charsiu/g2p_multilingual_byT5_tiny_12_layers_100/config.json

@@ -0,0 +1,32 @@
+{
+  "_name_or_path": "/home/patrick/t5/byt5-small",
+  "architectures": [
+    "T5ForConditionalGeneration"
+  ],
+  "d_ff": 1024,
+  "d_kv": 64,
+  "d_model": 256,
+  "decoder_start_token_id": 0,
+  "dense_act_fn": "gelu_new",
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-gelu",
+  "gradient_checkpointing": false,
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "is_gated_act": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "num_decoder_layers": 4,
+  "num_heads": 6,
+  "num_layers": 8,
+  "pad_token_id": 0,
+  "relative_attention_max_distance": 128,
+  "relative_attention_num_buckets": 32,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "ByT5Tokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.21.1",
+  "use_cache": true,
+  "vocab_size": 384
+}

byT5/Charsiu/g2p_multilingual_byT5_tiny_12_layers_100/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b759f8e71c6b67904a2cc43475f4c99cbd88cbb5724aca8bc74eccbc055b7eb
+size 63777973

byT5/Charsiu/g2p_multilingual_byT5_tiny_12_layers_100/source.txt

@@ -0,0 +1,2 @@
+https://huggingface.co/charsiu/g2p_multilingual_byT5_tiny_12_layers_100
+https://huggingface.co/Jarbas/charsiu_g2p_multilingual_byT5_tiny_12_layers_100_onnx

byT5/Charsiu/g2p_multilingual_byT5_tiny_12_layers_100/trainer_state.json

@@ -0,0 +1,1136 @@
+{
+  "best_metric": 0.05792941898107529,
+  "best_model_checkpoint": "/scratch/lingjzhu_root/lingjzhu1/lingjzhu/g2p/byt5_12_layers_baseline/checkpoint-135000",
+  "epoch": 9.917791237204689,
+  "global_step": 140000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003,
+      "loss": 7.2226,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002999623214094654,
+      "loss": 1.4006,
+      "step": 2000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002998493045668776,
+      "loss": 0.4781,
+      "step": 3000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029966100624977436,
+      "loss": 0.3172,
+      "step": 4000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029939752105569156,
+      "loss": 0.2595,
+      "step": 5000
+    },
+    {
+      "epoch": 0.35,
+      "eval_cer": 0.22637400547661657,
+      "eval_loss": 0.30229780077934265,
+      "eval_runtime": 79.8178,
+      "eval_samples_per_second": 62.643,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.5502,
+      "step": 5000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029905898135463893,
+      "loss": 0.2259,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002986455572226002,
+      "loss": 0.206,
+      "step": 7000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029815745635609,
+      "loss": 0.1889,
+      "step": 8000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029759492396781067,
+      "loss": 0.1758,
+      "step": 9000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00029695824266346256,
+      "loss": 0.1663,
+      "step": 10000
+    },
+    {
+      "epoch": 0.71,
+      "eval_cer": 0.17244873757519566,
+      "eval_loss": 0.19059504568576813,
+      "eval_runtime": 78.5038,
+      "eval_samples_per_second": 63.691,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.4426,
+      "step": 10000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000296247732299768,
+      "loss": 0.1572,
+      "step": 11000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000295463749823781,
+      "loss": 0.1501,
+      "step": 12000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002946066890935644,
+      "loss": 0.1429,
+      "step": 13000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002936769806803222,
+      "loss": 0.137,
+      "step": 14000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029267509165208945,
+      "loss": 0.1308,
+      "step": 15000
+    },
+    {
+      "epoch": 1.06,
+      "eval_cer": 0.15006791867008776,
+      "eval_loss": 0.15067286789417267,
+      "eval_runtime": 78.9896,
+      "eval_samples_per_second": 63.299,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.396,
+      "step": 15000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00029160152533908566,
+      "loss": 0.126,
+      "step": 16000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002904568210808509,
+      "loss": 0.1221,
+      "step": 17000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028924155395529237,
+      "loss": 0.1179,
+      "step": 18000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028795633448977547,
+      "loss": 0.1145,
+      "step": 19000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002866018083544069,
+      "loss": 0.1118,
+      "step": 20000
+    },
+    {
+      "epoch": 1.42,
+      "eval_cer": 0.1389853166303715,
+      "eval_loss": 0.1269478052854538,
+      "eval_runtime": 78.7424,
+      "eval_samples_per_second": 63.498,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.3746,
+      "step": 20000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002851786560376614,
+      "loss": 0.1081,
+      "step": 21000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00028368759250451786,
+      "loss": 0.1049,
+      "step": 22000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00028212936683727404,
+      "loss": 0.1025,
+      "step": 23000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002805047618592216,
+      "loss": 0.0997,
+      "step": 24000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002788145937413704,
+      "loss": 0.0982,
+      "step": 25000
+    },
+    {
+      "epoch": 1.77,
+      "eval_cer": 0.13158972810970482,
+      "eval_loss": 0.11525362730026245,
+      "eval_runtime": 79.1705,
+      "eval_samples_per_second": 63.155,
+      "eval_steps_per_second": 0.126,
+      "eval_wer": 0.3582,
+      "step": 25000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00027705971159241994,
+      "loss": 0.0962,
+      "step": 26000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002752409970321824,
+      "loss": 0.0936,
+      "step": 27000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000273359363748674,
+      "loss": 0.0929,
+      "step": 28000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002714157570390952,
+      "loss": 0.0893,
+      "step": 29000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002694111533349308,
+      "loss": 0.0876,
+      "step": 30000
+    },
+    {
+      "epoch": 2.13,
+      "eval_cer": 0.12440975441471355,
+      "eval_loss": 0.09987597167491913,
+      "eval_runtime": 78.8157,
+      "eval_samples_per_second": 63.439,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.3406,
+      "step": 30000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00026734655971140956,
+      "loss": 0.0863,
+      "step": 31000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026522301338156835,
+      "loss": 0.085,
+      "step": 32000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00026304158117517563,
+      "loss": 0.0838,
+      "step": 33000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002608033590027766,
+      "loss": 0.0826,
+      "step": 34000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000258509471305128,
+      "loss": 0.0812,
+      "step": 35000
+    },
+    {
+      "epoch": 2.48,
+      "eval_cer": 0.11850190819120723,
+      "eval_loss": 0.09199721366167068,
+      "eval_runtime": 78.9815,
+      "eval_samples_per_second": 63.306,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.3318,
+      "step": 35000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00025616107048830045,
+      "loss": 0.0799,
+      "step": 36000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002537593363447311,
+      "loss": 0.0791,
+      "step": 37000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002513054754605181,
+      "loss": 0.0784,
+      "step": 38000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002488007206092547,
+      "loss": 0.0771,
+      "step": 39000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00024624633013270664,
+      "loss": 0.0759,
+      "step": 40000
+    },
+    {
+      "epoch": 2.83,
+      "eval_cer": 0.11710041182431705,
+      "eval_loss": 0.08664757013320923,
+      "eval_runtime": 78.7555,
+      "eval_samples_per_second": 63.488,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.3278,
+      "step": 40000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002436435873086449,
+      "loss": 0.0753,
+      "step": 41000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00024099379970615137,
+      "loss": 0.0745,
+      "step": 42000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00023829829852872037,
+      "loss": 0.0724,
+      "step": 43000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00023555843794548724,
+      "loss": 0.0714,
+      "step": 44000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00023277559441091917,
+      "loss": 0.0706,
+      "step": 45000
+    },
+    {
+      "epoch": 3.19,
+      "eval_cer": 0.11453459539877962,
+      "eval_loss": 0.08330970257520676,
+      "eval_runtime": 79.5037,
+      "eval_samples_per_second": 62.89,
+      "eval_steps_per_second": 0.126,
+      "eval_wer": 0.324,
+      "step": 45000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002299511659733103,
+      "loss": 0.0699,
+      "step": 46000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002270865715724285,
+      "loss": 0.0694,
+      "step": 47000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0002241832503266669,
+      "loss": 0.0689,
+      "step": 48000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00022124266081005755,
+      "loss": 0.0683,
+      "step": 49000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0002182662803195115,
+      "loss": 0.0678,
+      "step": 50000
+    },
+    {
+      "epoch": 3.54,
+      "eval_cer": 0.11179628711270187,
+      "eval_loss": 0.07958581298589706,
+      "eval_runtime": 79.3907,
+      "eval_samples_per_second": 62.98,
+      "eval_steps_per_second": 0.126,
+      "eval_wer": 0.3198,
+      "step": 50000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00021525560413265246,
+      "loss": 0.0673,
+      "step": 51000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00021221214475661745,
+      "loss": 0.0662,
+      "step": 52000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020913743116820165,
+      "loss": 0.0661,
+      "step": 53000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020603300804572917,
+      "loss": 0.0653,
+      "step": 54000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00020290043499303573,
+      "loss": 0.065,
+      "step": 55000
+    },
+    {
+      "epoch": 3.9,
+      "eval_cer": 0.11030854481554152,
+      "eval_loss": 0.07400712370872498,
+      "eval_runtime": 78.8463,
+      "eval_samples_per_second": 63.414,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.3126,
+      "step": 55000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019974128575595291,
+      "loss": 0.064,
+      "step": 56000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.000196557147431688,
+      "loss": 0.0629,
+      "step": 57000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019334961967149597,
+      "loss": 0.062,
+      "step": 58000
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00019012031387704483,
+      "loss": 0.0612,
+      "step": 59000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018687085239087786,
+      "loss": 0.0612,
+      "step": 60000
+    },
+    {
+      "epoch": 4.25,
+      "eval_cer": 0.10946764699540741,
+      "eval_loss": 0.07309988141059875,
+      "eval_runtime": 78.4936,
+      "eval_samples_per_second": 63.7,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.3106,
+      "step": 60000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018360286768137894,
+      "loss": 0.0606,
+      "step": 61000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018031800152265137,
+      "loss": 0.0604,
+      "step": 62000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00017701790416972107,
+      "loss": 0.0596,
+      "step": 63000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017370423352948002,
+      "loss": 0.0598,
+      "step": 64000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017037865432778442,
+      "loss": 0.0591,
+      "step": 65000
+    },
+    {
+      "epoch": 4.6,
+      "eval_cer": 0.1068587076047349,
+      "eval_loss": 0.07157264649868011,
+      "eval_runtime": 78.9704,
+      "eval_samples_per_second": 63.315,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.3004,
+      "step": 65000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00016704283727312792,
+      "loss": 0.0585,
+      "step": 66000
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00016369845821730933,
+      "loss": 0.0585,
+      "step": 67000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016034719731351576,
+      "loss": 0.058,
+      "step": 68000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00015699073817224542,
+      "loss": 0.0577,
+      "step": 69000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00015363076701549341,
+      "loss": 0.0569,
+      "step": 70000
+    },
+    {
+      "epoch": 4.96,
+      "eval_cer": 0.10733306022122081,
+      "eval_loss": 0.0685366541147232,
+      "eval_runtime": 79.1043,
+      "eval_samples_per_second": 63.208,
+      "eval_steps_per_second": 0.126,
+      "eval_wer": 0.303,
+      "step": 70000
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00015026897182962535,
+      "loss": 0.0564,
+      "step": 71000
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00014690704151736484,
+      "loss": 0.0549,
+      "step": 72000
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0001435466650493204,
+      "loss": 0.0549,
+      "step": 73000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0001401895306154785,
+      "loss": 0.0545,
+      "step": 74000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0001368373247770885,
+      "loss": 0.0543,
+      "step": 75000
+    },
+    {
+      "epoch": 5.31,
+      "eval_cer": 0.10539252679014209,
+      "eval_loss": 0.06656810641288757,
+      "eval_runtime": 79.3205,
+      "eval_samples_per_second": 63.035,
+      "eval_steps_per_second": 0.126,
+      "eval_wer": 0.2982,
+      "step": 75000
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00013349173161936601,
+      "loss": 0.054,
+      "step": 76000
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00013015443190544018,
+      "loss": 0.0535,
+      "step": 77000
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00012682710223196978,
+      "loss": 0.0536,
+      "step": 78000
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00012351141418685273,
+      "loss": 0.0533,
+      "step": 79000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00012020903350945151,
+      "loss": 0.053,
+      "step": 80000
+    },
+    {
+      "epoch": 5.67,
+      "eval_cer": 0.10621186312770867,
+      "eval_loss": 0.0656103640794754,
+      "eval_runtime": 78.8232,
+      "eval_samples_per_second": 63.433,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.298,
+      "step": 80000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00011692161925375738,
+      "loss": 0.0527,
+      "step": 81000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00011365082295491233,
+      "loss": 0.0523,
+      "step": 82000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00011039828779950928,
+      "loss": 0.0522,
+      "step": 83000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0001071656478000858,
+      "loss": 0.052,
+      "step": 84000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00010395452697422703,
+      "loss": 0.0515,
+      "step": 85000
+    },
+    {
+      "epoch": 6.02,
+      "eval_cer": 0.1045085060048729,
+      "eval_loss": 0.06420928984880447,
+      "eval_runtime": 78.7185,
+      "eval_samples_per_second": 63.517,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.2958,
+      "step": 85000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00010076653852868962,
+      "loss": 0.05,
+      "step": 86000
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 9.760328404895728e-05,
+      "loss": 0.0498,
+      "step": 87000
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 9.446635269463389e-05,
+      "loss": 0.0493,
+      "step": 88000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 9.135732040107998e-05,
+      "loss": 0.0495,
+      "step": 89000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 8.82777490876921e-05,
+      "loss": 0.0494,
+      "step": 90000
+    },
+    {
+      "epoch": 6.38,
+      "eval_cer": 0.10474568231311585,
+      "eval_loss": 0.06237434595823288,
+      "eval_runtime": 79.1368,
+      "eval_samples_per_second": 63.182,
+      "eval_steps_per_second": 0.126,
+      "eval_wer": 0.2926,
+      "step": 90000
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 8.522918587322418e-05,
+      "loss": 0.0492,
+      "step": 91000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 8.22131622985438e-05,
+      "loss": 0.0492,
+      "step": 92000
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.923119355721526e-05,
+      "loss": 0.049,
+      "step": 93000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.628477773429438e-05,
+      "loss": 0.0486,
+      "step": 94000
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.3375395053719e-05,
+      "loss": 0.0486,
+      "step": 95000
+    },
+    {
+      "epoch": 6.73,
+      "eval_cer": 0.10392634597554928,
+      "eval_loss": 0.06150297820568085,
+      "eval_runtime": 78.9076,
+      "eval_samples_per_second": 63.365,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.2894,
+      "step": 95000
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 7.050450713467216e-05,
+      "loss": 0.0482,
+      "step": 96000
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.767355625729215e-05,
+      "loss": 0.0479,
+      "step": 97000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.488396463809805e-05,
+      "loss": 0.0477,
+      "step": 98000
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.213713371549497e-05,
+      "loss": 0.0473,
+      "step": 99000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.943444344571755e-05,
+      "loss": 0.0462,
+      "step": 100000
+    },
+    {
+      "epoch": 7.08,
+      "eval_cer": 0.10289139481230729,
+      "eval_loss": 0.06110595166683197,
+      "eval_runtime": 78.8023,
+      "eval_samples_per_second": 63.45,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.289,
+      "step": 100000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.6777251609566e-05,
+      "loss": 0.0463,
+      "step": 101000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.416689313028253e-05,
+      "loss": 0.046,
+      "step": 102000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.1604679402911015e-05,
+      "loss": 0.046,
+      "step": 103000
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.9091897635476726e-05,
+      "loss": 0.0459,
+      "step": 104000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.6629810202317166e-05,
+      "loss": 0.0459,
+      "step": 105000
+    },
+    {
+      "epoch": 7.44,
+      "eval_cer": 0.10291295629487483,
+      "eval_loss": 0.05981627106666565,
+      "eval_runtime": 78.8059,
+      "eval_samples_per_second": 63.447,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.289,
+      "step": 105000
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.421965400988918e-05,
+      "loss": 0.0457,
+      "step": 106000
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.1862639875370016e-05,
+      "loss": 0.0458,
+      "step": 107000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.955995191836581e-05,
+      "loss": 0.0456,
+      "step": 108000
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.731274696603201e-05,
+      "loss": 0.0454,
+      "step": 109000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.512215397190527e-05,
+      "loss": 0.0451,
+      "step": 110000
+    },
+    {
+      "epoch": 7.79,
+      "eval_cer": 0.10304232519028009,
+      "eval_loss": 0.05932984501123428,
+      "eval_runtime": 79.019,
+      "eval_samples_per_second": 63.276,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.2862,
+      "step": 110000
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.298927344873821e-05,
+      "loss": 0.045,
+      "step": 111000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.091517691562284e-05,
+      "loss": 0.0448,
+      "step": 112000
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.890090635967912e-05,
+      "loss": 0.0454,
+      "step": 113000
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.694747371258042e-05,
+      "loss": 0.0441,
+      "step": 114000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.505586034217802e-05,
+      "loss": 0.0436,
+      "step": 115000
+    },
+    {
+      "epoch": 8.15,
+      "eval_cer": 0.10248172664352401,
+      "eval_loss": 0.0589553639292717,
+      "eval_runtime": 79.3196,
+      "eval_samples_per_second": 63.036,
+      "eval_steps_per_second": 0.126,
+      "eval_wer": 0.2848,
+      "step": 115000
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.3227016559480356e-05,
+      "loss": 0.0438,
+      "step": 116000
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.1461861141234628e-05,
+      "loss": 0.044,
+      "step": 117000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.9761280868350654e-05,
+      "loss": 0.0437,
+      "step": 118000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.8126130080398725e-05,
+      "loss": 0.0438,
+      "step": 119000
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.655723024640553e-05,
+      "loss": 0.0435,
+      "step": 120000
+    },
+    {
+      "epoch": 8.5,
+      "eval_cer": 0.10250328812609155,
+      "eval_loss": 0.05871221795678139,
+      "eval_runtime": 79.0526,
+      "eval_samples_per_second": 63.249,
+      "eval_steps_per_second": 0.126,
+      "eval_wer": 0.2844,
+      "step": 120000
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.5055369552163509e-05,
+      "loss": 0.0435,
+      "step": 121000
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.3621302504261194e-05,
+      "loss": 0.0436,
+      "step": 122000
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.2255749551033205e-05,
+      "loss": 0.0435,
+      "step": 123000
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.0959396720620555e-05,
+      "loss": 0.0435,
+      "step": 124000
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 9.732895276323099e-06,
+      "loss": 0.0434,
+      "step": 125000
+    },
+    {
+      "epoch": 8.86,
+      "eval_cer": 0.10222298885271351,
+      "eval_loss": 0.058125074952840805,
+      "eval_runtime": 78.8946,
+      "eval_samples_per_second": 63.376,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.2842,
+      "step": 125000
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 8.576861389416939e-06,
+      "loss": 0.0433,
+      "step": 126000
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.4918758296016734e-06,
+      "loss": 0.0431,
+      "step": 127000
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 6.478483673232537e-06,
+      "loss": 0.0431,
+      "step": 128000
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.537194029484349e-06,
+      "loss": 0.0427,
+      "step": 129000
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.668479784584583e-06,
+      "loss": 0.0424,
+      "step": 130000
+    },
+    {
+      "epoch": 9.21,
+      "eval_cer": 0.1017486362362276,
+      "eval_loss": 0.05813354626297951,
+      "eval_runtime": 78.9118,
+      "eval_samples_per_second": 63.362,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.283,
+      "step": 130000
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.872777364244245e-06,
+      "loss": 0.0429,
+      "step": 131000
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.150486514405798e-06,
+      "loss": 0.0426,
+      "step": 132000
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.50197010041826e-06,
+      "loss": 0.0425,
+      "step": 133000
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.9275539247405437e-06,
+      "loss": 0.0426,
+      "step": 134000
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4275265632643806e-06,
+      "loss": 0.0429,
+      "step": 135000
+    },
+    {
+      "epoch": 9.56,
+      "eval_cer": 0.10192112809676794,
+      "eval_loss": 0.05792941898107529,
+      "eval_runtime": 78.9437,
+      "eval_samples_per_second": 63.336,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.2832,
+      "step": 135000
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.0021392203391932e-06,
+      "loss": 0.0423,
+      "step": 136000
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 6.516056025718585e-07,
+      "loss": 0.0427,
+      "step": 137000
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.76101811464391e-07,
+      "loss": 0.0428,
+      "step": 138000
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.7576625494392648e-07,
+      "loss": 0.0428,
+      "step": 139000
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.0699577829210744e-08,
+      "loss": 0.0428,
+      "step": 140000
+    },
+    {
+      "epoch": 9.92,
+      "eval_cer": 0.10161926734082236,
+      "eval_loss": 0.05793117731809616,
+      "eval_runtime": 79.0415,
+      "eval_samples_per_second": 63.258,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.2822,
+      "step": 140000
+    }
+  ],
+  "max_steps": 141160,
+  "num_train_epochs": 10,
+  "total_flos": 2.733854879374295e+17,
+  "trial_name": null,
+  "trial_params": null
+}

byT5/Charsiu/g2p_multilingual_byT5_tiny_16_layers/.gitattributes

@@ -0,0 +1,27 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

byT5/Charsiu/g2p_multilingual_byT5_tiny_16_layers/config.json

@@ -0,0 +1,29 @@
+{
+  "_name_or_path": "/home/patrick/t5/byt5-small",
+  "architectures": [
+    "T5ForConditionalGeneration"
+  ],
+  "d_ff": 1024,
+  "d_kv": 64,
+  "d_model": 256,
+  "decoder_start_token_id": 0,
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-gelu",
+  "gradient_checkpointing": false,
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "num_decoder_layers": 4,
+  "num_heads": 6,
+  "num_layers": 12,
+  "pad_token_id": 0,
+  "relative_attention_num_buckets": 32,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "ByT5Tokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.16.2",
+  "use_cache": true,
+  "vocab_size": 384
+}

byT5/Charsiu/g2p_multilingual_byT5_tiny_16_layers/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a3699cd4ccf4ca53e3690cb001828bbd7cf1c583f474edf7e8e77995f1065fa
+size 82692269

byT5/Charsiu/g2p_multilingual_byT5_tiny_16_layers/source.txt

@@ -0,0 +1 @@
+https://huggingface.co/charsiu/g2p_multilingual_byT5_tiny_16_layers

byT5/Charsiu/g2p_multilingual_byT5_tiny_16_layers/trainer_state.json

@@ -0,0 +1,1096 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.646302250803858,
+  "global_step": 135000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003,
+      "loss": 2.0973,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002999616623572683,
+      "loss": 0.5502,
+      "step": 2000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029984666902607135,
+      "loss": 0.2896,
+      "step": 3000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002996550787873857,
+      "loss": 0.2299,
+      "step": 4000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002993869895761197,
+      "loss": 0.2005,
+      "step": 5000
+    },
+    {
+      "epoch": 0.36,
+      "eval_cer": 0.2034279967643908,
+      "eval_loss": 0.2542494535446167,
+      "eval_runtime": 84.4226,
+      "eval_samples_per_second": 58.634,
+      "eval_steps_per_second": 0.118,
+      "eval_wer": 0.5028282828282828,
+      "step": 5000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000299042538431052,
+      "loss": 0.1818,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029862190142478177,
+      "loss": 0.168,
+      "step": 7000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029812529357372587,
+      "loss": 0.1575,
+      "step": 8000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029755296872820933,
+      "loss": 0.1488,
+      "step": 9000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002969052194427048,
+      "loss": 0.1409,
+      "step": 10000
+    },
+    {
+      "epoch": 0.71,
+      "eval_cer": 0.15823877921339718,
+      "eval_loss": 0.17229020595550537,
+      "eval_runtime": 83.5893,
+      "eval_samples_per_second": 59.218,
+      "eval_steps_per_second": 0.12,
+      "eval_wer": 0.41454545454545455,
+      "step": 10000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002961823768262882,
+      "loss": 0.1342,
+      "step": 11000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002953848103733858,
+      "loss": 0.128,
+      "step": 12000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029451292777490066,
+      "loss": 0.1221,
+      "step": 13000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002935671747098137,
+      "loss": 0.1175,
+      "step": 14000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029254803461736643,
+      "loss": 0.1114,
+      "step": 15000
+    },
+    {
+      "epoch": 1.07,
+      "eval_cer": 0.1419514221376883,
+      "eval_loss": 0.13527829945087433,
+      "eval_runtime": 83.6072,
+      "eval_samples_per_second": 59.205,
+      "eval_steps_per_second": 0.12,
+      "eval_wer": 0.37757575757575756,
+      "step": 15000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00029145602844994243,
+      "loss": 0.1085,
+      "step": 16000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002902917144067724,
+      "loss": 0.1043,
+      "step": 17000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028905568764860047,
+      "loss": 0.1012,
+      "step": 18000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028774857999345685,
+      "loss": 0.099,
+      "step": 19000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002863710595936922,
+      "loss": 0.096,
+      "step": 20000
+    },
+    {
+      "epoch": 1.43,
+      "eval_cer": 0.13088913666076388,
+      "eval_loss": 0.11353930085897446,
+      "eval_runtime": 83.6733,
+      "eval_samples_per_second": 59.159,
+      "eval_steps_per_second": 0.12,
+      "eval_wer": 0.35434343434343435,
+      "step": 20000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002849238305944389,
+      "loss": 0.0934,
+      "step": 21000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00028340763277367477,
+      "loss": 0.0916,
+      "step": 22000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002818232411640713,
+      "loss": 0.0891,
+      "step": 23000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00028017146565682144,
+      "loss": 0.0882,
+      "step": 24000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00027845315058764886,
+      "loss": 0.0852,
+      "step": 25000
+    },
+    {
+      "epoch": 1.79,
+      "eval_cer": 0.12312804704750661,
+      "eval_loss": 0.10132956504821777,
+      "eval_runtime": 83.2401,
+      "eval_samples_per_second": 59.467,
+      "eval_steps_per_second": 0.12,
+      "eval_wer": 0.3373737373737374,
+      "step": 25000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00027666917430520975,
+      "loss": 0.0836,
+      "step": 26000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00027482044872210895,
+      "loss": 0.0819,
+      "step": 27000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002729079188487587,
+      "loss": 0.0806,
+      "step": 28000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00027093256231031885,
+      "loss": 0.0773,
+      "step": 29000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00026889538884696597,
+      "loss": 0.0769,
+      "step": 30000
+    },
+    {
+      "epoch": 2.14,
+      "eval_cer": 0.1182309088126626,
+      "eval_loss": 0.09128542244434357,
+      "eval_runtime": 83.8171,
+      "eval_samples_per_second": 59.057,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.3333333333333333,
+      "step": 30000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002667974397977457,
+      "loss": 0.0757,
+      "step": 31000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002646397875682729,
+      "loss": 0.0745,
+      "step": 32000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00026242353508255185,
+      "loss": 0.0737,
+      "step": 33000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002601498152191957,
+      "loss": 0.073,
+      "step": 34000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002578197902323352,
+      "loss": 0.072,
+      "step": 35000
+    },
+    {
+      "epoch": 2.5,
+      "eval_cer": 0.11523578408867319,
+      "eval_loss": 0.0870400071144104,
+      "eval_runtime": 83.456,
+      "eval_samples_per_second": 59.313,
+      "eval_steps_per_second": 0.12,
+      "eval_wer": 0.32666666666666666,
+      "step": 35000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00025543465115751026,
+      "loss": 0.0707,
+      "step": 36000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0002529956172028505,
+      "loss": 0.07,
+      "step": 37000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002505039351258541,
+      "loss": 0.0692,
+      "step": 38000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002479608785960846,
+      "loss": 0.0682,
+      "step": 39000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0002453677475441111,
+      "loss": 0.0672,
+      "step": 40000
+    },
+    {
+      "epoch": 2.86,
+      "eval_cer": 0.11036050807809186,
+      "eval_loss": 0.08000758290290833,
+      "eval_runtime": 84.1634,
+      "eval_samples_per_second": 58.814,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.31292929292929295,
+      "step": 40000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00024272586749702474,
+      "loss": 0.0666,
+      "step": 41000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002400365889008706,
+      "loss": 0.0662,
+      "step": 42000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023730128643034235,
+      "loss": 0.0631,
+      "step": 43000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00023452135828609167,
+      "loss": 0.0635,
+      "step": 44000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0002316982254800121,
+      "loss": 0.0625,
+      "step": 45000
+    },
+    {
+      "epoch": 3.22,
+      "eval_cer": 0.10987953914431255,
+      "eval_loss": 0.0780956819653511,
+      "eval_runtime": 83.8176,
+      "eval_samples_per_second": 59.057,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.30646464646464644,
+      "step": 45000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00022883333110886237,
+      "loss": 0.0622,
+      "step": 46000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00022592813961660067,
+      "loss": 0.0615,
+      "step": 47000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00022298413604580696,
+      "loss": 0.0608,
+      "step": 48000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00022000282527857588,
+      "loss": 0.0604,
+      "step": 49000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0002169857312672683,
+      "loss": 0.0597,
+      "step": 50000
+    },
+    {
+      "epoch": 3.57,
+      "eval_cer": 0.108261734548873,
+      "eval_loss": 0.07320722192525864,
+      "eval_runtime": 84.043,
+      "eval_samples_per_second": 58.898,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.30484848484848487,
+      "step": 50000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00021393439625551483,
+      "loss": 0.0598,
+      "step": 51000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00021085037998986924,
+      "loss": 0.0588,
+      "step": 52000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020773525892251514,
+      "loss": 0.0587,
+      "step": 53000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00020459062540543316,
+      "loss": 0.0582,
+      "step": 54000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00020141808687644067,
+      "loss": 0.0572,
+      "step": 55000
+    },
+    {
+      "epoch": 3.93,
+      "eval_cer": 0.10843663234297457,
+      "eval_loss": 0.0711025819182396,
+      "eval_runtime": 83.6072,
+      "eval_samples_per_second": 59.205,
+      "eval_steps_per_second": 0.12,
+      "eval_wer": 0.30262626262626263,
+      "step": 55000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019821926503751995,
+      "loss": 0.0568,
+      "step": 56000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019499579502585537,
+      "loss": 0.0549,
+      "step": 57000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019174932457800242,
+      "loss": 0.0544,
+      "step": 58000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001884815131876167,
+      "loss": 0.0542,
+      "step": 59000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018519403125717278,
+      "loss": 0.0537,
+      "step": 60000
+    },
+    {
+      "epoch": 4.29,
+      "eval_cer": 0.10559454318882402,
+      "eval_loss": 0.06793416291475296,
+      "eval_runtime": 83.7622,
+      "eval_samples_per_second": 59.096,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.29474747474747476,
+      "step": 60000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018188855924410722,
+      "loss": 0.0539,
+      "step": 61000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00017856678680182127,
+      "loss": 0.0534,
+      "step": 62000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001752304119159834,
+      "loss": 0.053,
+      "step": 63000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017188114003657205,
+      "loss": 0.0528,
+      "step": 64000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00016852068320610358,
+      "loss": 0.052,
+      "step": 65000
+    },
+    {
+      "epoch": 4.64,
+      "eval_cer": 0.10421722306027416,
+      "eval_loss": 0.06535057723522186,
+      "eval_runtime": 83.4596,
+      "eval_samples_per_second": 59.31,
+      "eval_steps_per_second": 0.12,
+      "eval_wer": 0.2911111111111111,
+      "step": 65000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00016515075918448972,
+      "loss": 0.0519,
+      "step": 66000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016177309057097285,
+      "loss": 0.0511,
+      "step": 67000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00015838940392358722,
+      "loss": 0.0512,
+      "step": 68000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00015500142887659688,
+      "loss": 0.0506,
+      "step": 69000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00015161089725636095,
+      "loss": 0.0506,
+      "step": 70000
+    },
+    {
+      "epoch": 5.0,
+      "eval_cer": 0.10323342296845281,
+      "eval_loss": 0.06469610333442688,
+      "eval_runtime": 83.521,
+      "eval_samples_per_second": 59.267,
+      "eval_steps_per_second": 0.12,
+      "eval_wer": 0.2909090909090909,
+      "step": 70000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00014821954219607845,
+      "loss": 0.0483,
+      "step": 71000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0001448290972498651,
+      "loss": 0.0483,
+      "step": 72000
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00014144129550661485,
+      "loss": 0.048,
+      "step": 73000
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0001380578687040995,
+      "loss": 0.0478,
+      "step": 74000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00013468054634375843,
+      "loss": 0.0481,
+      "step": 75000
+    },
+    {
+      "epoch": 5.36,
+      "eval_cer": 0.10148444502743709,
+      "eval_loss": 0.06252569705247879,
+      "eval_runtime": 83.6066,
+      "eval_samples_per_second": 59.206,
+      "eval_steps_per_second": 0.12,
+      "eval_wer": 0.2886868686868687,
+      "step": 75000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00013131105480663235,
+      "loss": 0.0476,
+      "step": 76000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.000127951116470891,
+      "loss": 0.0472,
+      "step": 77000
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00012460244883140783,
+      "loss": 0.0469,
+      "step": 78000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0001212667636218309,
+      "loss": 0.0465,
+      "step": 79000
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00011794576593959775,
+      "loss": 0.0462,
+      "step": 80000
+    },
+    {
+      "epoch": 5.72,
+      "eval_cer": 0.10179051616711485,
+      "eval_loss": 0.06165655702352524,
+      "eval_runtime": 83.7971,
+      "eval_samples_per_second": 59.071,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.28383838383838383,
+      "step": 80000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00011464115337434394,
+      "loss": 0.0463,
+      "step": 81000
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00011135461514014796,
+      "loss": 0.0456,
+      "step": 82000
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00010808783121205837,
+      "loss": 0.0456,
+      "step": 83000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00010484247146734352,
+      "loss": 0.0454,
+      "step": 84000
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00010162019483190237,
+      "loss": 0.0433,
+      "step": 85000
+    },
+    {
+      "epoch": 6.07,
+      "eval_cer": 0.10122209833628473,
+      "eval_loss": 0.05922077223658562,
+      "eval_runtime": 83.6915,
+      "eval_samples_per_second": 59.146,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.2824242424242424,
+      "step": 85000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 9.842264843227404e-05,
+      "loss": 0.0439,
+      "step": 86000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 9.52514667536784e-05,
+      "loss": 0.043,
+      "step": 87000
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 9.210827080451842e-05,
+      "loss": 0.0431,
+      "step": 88000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 8.899466728777203e-05,
+      "loss": 0.0434,
+      "step": 89000
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 8.591224777969557e-05,
+      "loss": 0.0428,
+      "step": 90000
+    },
+    {
+      "epoch": 6.43,
+      "eval_cer": 0.0995168448437944,
+      "eval_loss": 0.05855976790189743,
+      "eval_runtime": 83.8689,
+      "eval_samples_per_second": 59.021,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.2791919191919192,
+      "step": 90000
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 8.286258791626041e-05,
+      "loss": 0.0428,
+      "step": 91000
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.984724658773716e-05,
+      "loss": 0.0427,
+      "step": 92000
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.686776514184009e-05,
+      "loss": 0.0424,
+      "step": 93000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.392566659583846e-05,
+      "loss": 0.0422,
+      "step": 94000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 7.102245485803813e-05,
+      "loss": 0.0421,
+      "step": 95000
+    },
+    {
+      "epoch": 6.79,
+      "eval_cer": 0.10063181828119193,
+      "eval_loss": 0.05791032314300537,
+      "eval_runtime": 83.4355,
+      "eval_samples_per_second": 59.327,
+      "eval_steps_per_second": 0.12,
+      "eval_wer": 0.2822222222222222,
+      "step": 95000
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.81596139590308e-05,
+      "loss": 0.042,
+      "step": 96000
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.533860729310434e-05,
+      "loss": 0.0416,
+      "step": 97000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.256087687020127e-05,
+      "loss": 0.0416,
+      "step": 98000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.98278425788092e-05,
+      "loss": 0.0402,
+      "step": 99000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.71409014601578e-05,
+      "loss": 0.0399,
+      "step": 100000
+    },
+    {
+      "epoch": 7.15,
+      "eval_cer": 0.09980105375920946,
+      "eval_loss": 0.05731820687651634,
+      "eval_runtime": 83.6985,
+      "eval_samples_per_second": 59.141,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.2775757575757576,
+      "step": 100000
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.4501426994095876e-05,
+      "loss": 0.0397,
+      "step": 101000
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.191076839701103e-05,
+      "loss": 0.0397,
+      "step": 102000
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.9370249932153075e-05,
+      "loss": 0.0398,
+      "step": 103000
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.6881170232712164e-05,
+      "loss": 0.0397,
+      "step": 104000
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.444480163799822e-05,
+      "loss": 0.0396,
+      "step": 105000
+    },
+    {
+      "epoch": 7.5,
+      "eval_cer": 0.0988828403401762,
+      "eval_loss": 0.05615651234984398,
+      "eval_runtime": 84.0673,
+      "eval_samples_per_second": 58.881,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.27656565656565657,
+      "step": 105000
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.2062389543061265e-05,
+      "loss": 0.0395,
+      "step": 106000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.9735151762084384e-05,
+      "loss": 0.0394,
+      "step": 107000
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.746427790587557e-05,
+      "loss": 0.0389,
+      "step": 108000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.525092877377602e-05,
+      "loss": 0.0392,
+      "step": 109000
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.309623576029597e-05,
+      "loss": 0.0388,
+      "step": 110000
+    },
+    {
+      "epoch": 7.86,
+      "eval_cer": 0.09962615596510789,
+      "eval_loss": 0.05532450973987579,
+      "eval_runtime": 83.843,
+      "eval_samples_per_second": 59.039,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.281010101010101,
+      "step": 110000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.1001300276781274e-05,
+      "loss": 0.039,
+      "step": 111000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.8967193188406938e-05,
+      "loss": 0.0386,
+      "step": 112000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.699495426678389e-05,
+      "loss": 0.0379,
+      "step": 113000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.5085591658461056e-05,
+      "loss": 0.0376,
+      "step": 114000
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.3240081369591984e-05,
+      "loss": 0.0375,
+      "step": 115000
+    },
+    {
+      "epoch": 8.22,
+      "eval_cer": 0.09870794254607464,
+      "eval_loss": 0.055045340210199356,
+      "eval_runtime": 83.5002,
+      "eval_samples_per_second": 59.281,
+      "eval_steps_per_second": 0.12,
+      "eval_wer": 0.27636363636363637,
+      "step": 115000
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.1459366767031522e-05,
+      "loss": 0.0377,
+      "step": 116000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.9744358096116225e-05,
+      "loss": 0.0375,
+      "step": 117000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.8095932015375496e-05,
+      "loss": 0.0374,
+      "step": 118000
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.65149311484114e-05,
+      "loss": 0.0373,
+      "step": 119000
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.500216365317587e-05,
+      "loss": 0.0372,
+      "step": 120000
+    },
+    {
+      "epoch": 8.57,
+      "eval_cer": 0.09803021359393105,
+      "eval_loss": 0.05470244958996773,
+      "eval_runtime": 83.9735,
+      "eval_samples_per_second": 58.947,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.27454545454545454,
+      "step": 120000
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.355840280886582e-05,
+      "loss": 0.0374,
+      "step": 121000
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.2184386620647097e-05,
+      "loss": 0.0375,
+      "step": 122000
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.0880817442409478e-05,
+      "loss": 0.0374,
+      "step": 123000
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 9.648361617745371e-06,
+      "loss": 0.0372,
+      "step": 124000
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 8.487649139335962e-06,
+      "loss": 0.0373,
+      "step": 125000
+    },
+    {
+      "epoch": 8.93,
+      "eval_cer": 0.09853304475197307,
+      "eval_loss": 0.054629139602184296,
+      "eval_runtime": 83.9435,
+      "eval_samples_per_second": 58.968,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.27575757575757576,
+      "step": 125000
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.399273326918692e-06,
+      "loss": 0.0372,
+      "step": 126000
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 6.383790524001009e-06,
+      "loss": 0.0367,
+      "step": 127000
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.441719813474849e-06,
+      "loss": 0.0367,
+      "step": 128000
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.57354275227797e-06,
+      "loss": 0.0366,
+      "step": 129000
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.7797031252369767e-06,
+      "loss": 0.0363,
+      "step": 130000
+    },
+    {
+      "epoch": 9.29,
+      "eval_cer": 0.09787717802409217,
+      "eval_loss": 0.05455449968576431,
+      "eval_runtime": 83.6734,
+      "eval_samples_per_second": 59.159,
+      "eval_steps_per_second": 0.12,
+      "eval_wer": 0.2739393939393939,
+      "step": 130000
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.0606067182186776e-06,
+      "loss": 0.0362,
+      "step": 131000
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.4166211107049584e-06,
+      "loss": 0.0365,
+      "step": 132000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8480754878977489e-06,
+      "loss": 0.0365,
+      "step": 133000
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3552604724498928e-06,
+      "loss": 0.0366,
+      "step": 134000
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 9.384279759080127e-07,
+      "loss": 0.0366,
+      "step": 135000
+    },
+    {
+      "epoch": 9.65,
+      "eval_cer": 0.09768041800572791,
+      "eval_loss": 0.05442119389772415,
+      "eval_runtime": 83.7582,
+      "eval_samples_per_second": 59.099,
+      "eval_steps_per_second": 0.119,
+      "eval_wer": 0.2719191919191919,
+      "step": 135000
+    }
+  ],
+  "max_steps": 139950,
+  "num_train_epochs": 10,
+  "total_flos": 3.424533742064886e+17,
+  "trial_name": null,
+  "trial_params": null
+}

byT5/Charsiu/g2p_multilingual_byT5_tiny_16_layers_100/.gitattributes

@@ -0,0 +1,31 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

byT5/Charsiu/g2p_multilingual_byT5_tiny_16_layers_100/charsiu_g2p_multilingual_byT5_tiny_16_layers_100.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2d602a71a0acff018a675538b5ac78619fccdd30dcc17c9d0d411f25f46c53c
+size 82979227

byT5/Charsiu/g2p_multilingual_byT5_tiny_16_layers_100/config.json

@@ -0,0 +1,32 @@
+{
+  "_name_or_path": "/home/patrick/t5/byt5-small",
+  "architectures": [
+    "T5ForConditionalGeneration"
+  ],
+  "d_ff": 1024,
+  "d_kv": 64,
+  "d_model": 256,
+  "decoder_start_token_id": 0,
+  "dense_act_fn": "gelu_new",
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-gelu",
+  "gradient_checkpointing": false,
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "is_gated_act": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "num_decoder_layers": 4,
+  "num_heads": 6,
+  "num_layers": 12,
+  "pad_token_id": 0,
+  "relative_attention_max_distance": 128,
+  "relative_attention_num_buckets": 32,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "ByT5Tokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.21.1",
+  "use_cache": true,
+  "vocab_size": 384
+}

byT5/Charsiu/g2p_multilingual_byT5_tiny_16_layers_100/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1fdf8971dd0386d1d4a0da95d538a7cce1477cf2754587f818e5ba039a7cf130
+size 82672813

byT5/Charsiu/g2p_multilingual_byT5_tiny_16_layers_100/source.txt

@@ -0,0 +1,2 @@
+https://huggingface.co/charsiu/g2p_multilingual_byT5_tiny_16_layers_100
+https://huggingface.co/Jarbas/charsiu_g2p_multilingual_byT5_tiny_16_layers_100_onnx

byT5/Charsiu/g2p_multilingual_byT5_tiny_16_layers_100/trainer_state.json

@@ -0,0 +1,1136 @@
+{
+  "best_metric": 0.05459373816847801,
+  "best_model_checkpoint": "/scratch/lingjzhu_root/lingjzhu1/lingjzhu/g2p/byt5_16_layers_baseline/checkpoint-140000",
+  "epoch": 9.917791237204689,
+  "global_step": 140000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003,
+      "loss": 7.0422,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002999623214094654,
+      "loss": 1.3957,
+      "step": 2000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002998493045668776,
+      "loss": 0.4792,
+      "step": 3000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029966100624977436,
+      "loss": 0.304,
+      "step": 4000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029939752105569156,
+      "loss": 0.2449,
+      "step": 5000
+    },
+    {
+      "epoch": 0.35,
+      "eval_cer": 0.21979775329351647,
+      "eval_loss": 0.2839755415916443,
+      "eval_runtime": 80.4341,
+      "eval_samples_per_second": 62.163,
+      "eval_steps_per_second": 0.124,
+      "eval_wer": 0.5296,
+      "step": 5000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029905898135463893,
+      "loss": 0.2141,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002986455572226002,
+      "loss": 0.1944,
+      "step": 7000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029815745635609,
+      "loss": 0.1791,
+      "step": 8000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029759492396781067,
+      "loss": 0.1669,
+      "step": 9000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00029695824266346256,
+      "loss": 0.1574,
+      "step": 10000
+    },
+    {
+      "epoch": 0.71,
+      "eval_cer": 0.16565687056642014,
+      "eval_loss": 0.1845117211341858,
+      "eval_runtime": 79.5791,
+      "eval_samples_per_second": 62.831,
+      "eval_steps_per_second": 0.126,
+      "eval_wer": 0.4346,
+      "step": 10000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000296247732299768,
+      "loss": 0.1486,
+      "step": 11000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000295463749823781,
+      "loss": 0.1411,
+      "step": 12000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002946066890935644,
+      "loss": 0.1339,
+      "step": 13000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002936769806803222,
+      "loss": 0.1287,
+      "step": 14000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029267509165208945,
+      "loss": 0.1215,
+      "step": 15000
+    },
+    {
+      "epoch": 1.06,
+      "eval_cer": 0.14532439250522866,
+      "eval_loss": 0.1390569657087326,
+      "eval_runtime": 79.7847,
+      "eval_samples_per_second": 62.669,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.3858,
+      "step": 15000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00029160152533908566,
+      "loss": 0.1181,
+      "step": 16000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002904568210808509,
+      "loss": 0.1137,
+      "step": 17000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028924155395529237,
+      "loss": 0.11,
+      "step": 18000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028795633448977547,
+      "loss": 0.1067,
+      "step": 19000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002866018083544069,
+      "loss": 0.1042,
+      "step": 20000
+    },
+    {
+      "epoch": 1.42,
+      "eval_cer": 0.1343495978783501,
+      "eval_loss": 0.11732806265354156,
+      "eval_runtime": 79.7707,
+      "eval_samples_per_second": 62.68,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.3624,
+      "step": 20000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002851786560376614,
+      "loss": 0.101,
+      "step": 21000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00028368759250451786,
+      "loss": 0.0983,
+      "step": 22000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00028212936683727404,
+      "loss": 0.0954,
+      "step": 23000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002805047618592216,
+      "loss": 0.0933,
+      "step": 24000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002788145937413704,
+      "loss": 0.0917,
+      "step": 25000
+    },
+    {
+      "epoch": 1.77,
+      "eval_cer": 0.1251644063045775,
+      "eval_loss": 0.10654760897159576,
+      "eval_runtime": 79.786,
+      "eval_samples_per_second": 62.668,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.3488,
+      "step": 25000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00027705971159241994,
+      "loss": 0.0898,
+      "step": 26000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002752409970321824,
+      "loss": 0.0881,
+      "step": 27000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000273359363748674,
+      "loss": 0.0865,
+      "step": 28000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002714157570390952,
+      "loss": 0.0834,
+      "step": 29000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002694111533349308,
+      "loss": 0.0821,
+      "step": 30000
+    },
+    {
+      "epoch": 2.13,
+      "eval_cer": 0.11837253929580198,
+      "eval_loss": 0.09495185315608978,
+      "eval_runtime": 79.9216,
+      "eval_samples_per_second": 62.561,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.3354,
+      "step": 30000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00026734655971140956,
+      "loss": 0.0806,
+      "step": 31000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026522301338156835,
+      "loss": 0.0791,
+      "step": 32000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00026304158117517563,
+      "loss": 0.0785,
+      "step": 33000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002608033590027766,
+      "loss": 0.0773,
+      "step": 34000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000258509471305128,
+      "loss": 0.0762,
+      "step": 35000
+    },
+    {
+      "epoch": 2.48,
+      "eval_cer": 0.1154617391491839,
+      "eval_loss": 0.08647703379392624,
+      "eval_runtime": 79.5904,
+      "eval_samples_per_second": 62.822,
+      "eval_steps_per_second": 0.126,
+      "eval_wer": 0.3272,
+      "step": 35000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00025616107048830045,
+      "loss": 0.0747,
+      "step": 36000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002537593363447311,
+      "loss": 0.0741,
+      "step": 37000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002513054754605181,
+      "loss": 0.0734,
+      "step": 38000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002488007206092547,
+      "loss": 0.0724,
+      "step": 39000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00024624633013270664,
+      "loss": 0.0715,
+      "step": 40000
+    },
+    {
+      "epoch": 2.83,
+      "eval_cer": 0.11341339830526748,
+      "eval_loss": 0.08288487792015076,
+      "eval_runtime": 79.5687,
+      "eval_samples_per_second": 62.839,
+      "eval_steps_per_second": 0.126,
+      "eval_wer": 0.3176,
+      "step": 40000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002436435873086449,
+      "loss": 0.0708,
+      "step": 41000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00024099379970615137,
+      "loss": 0.0695,
+      "step": 42000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00023829829852872037,
+      "loss": 0.0677,
+      "step": 43000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00023555843794548724,
+      "loss": 0.0666,
+      "step": 44000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00023277559441091917,
+      "loss": 0.0661,
+      "step": 45000
+    },
+    {
+      "epoch": 3.19,
+      "eval_cer": 0.11099851225770284,
+      "eval_loss": 0.07811947166919708,
+      "eval_runtime": 79.5629,
+      "eval_samples_per_second": 62.843,
+      "eval_steps_per_second": 0.126,
+      "eval_wer": 0.316,
+      "step": 45000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002299511659733103,
+      "loss": 0.0655,
+      "step": 46000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002270865715724285,
+      "loss": 0.0651,
+      "step": 47000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0002241832503266669,
+      "loss": 0.0645,
+      "step": 48000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00022124266081005755,
+      "loss": 0.0641,
+      "step": 49000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0002182662803195115,
+      "loss": 0.0636,
+      "step": 50000
+    },
+    {
+      "epoch": 3.54,
+      "eval_cer": 0.10931671661743461,
+      "eval_loss": 0.07468874752521515,
+      "eval_runtime": 80.209,
+      "eval_samples_per_second": 62.337,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.3074,
+      "step": 50000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00021525560413265246,
+      "loss": 0.0629,
+      "step": 51000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00021221214475661745,
+      "loss": 0.0619,
+      "step": 52000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020913743116820165,
+      "loss": 0.0619,
+      "step": 53000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020603300804572917,
+      "loss": 0.0608,
+      "step": 54000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00020290043499303573,
+      "loss": 0.061,
+      "step": 55000
+    },
+    {
+      "epoch": 3.9,
+      "eval_cer": 0.11015761443756873,
+      "eval_loss": 0.07217952609062195,
+      "eval_runtime": 79.9934,
+      "eval_samples_per_second": 62.505,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.3086,
+      "step": 55000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019974128575595291,
+      "loss": 0.06,
+      "step": 56000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.000196557147431688,
+      "loss": 0.0585,
+      "step": 57000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019334961967149597,
+      "loss": 0.0577,
+      "step": 58000
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00019012031387704483,
+      "loss": 0.0574,
+      "step": 59000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018687085239087786,
+      "loss": 0.0569,
+      "step": 60000
+    },
+    {
+      "epoch": 4.25,
+      "eval_cer": 0.10580219495892537,
+      "eval_loss": 0.06902794539928436,
+      "eval_runtime": 79.9992,
+      "eval_samples_per_second": 62.501,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.2988,
+      "step": 60000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018360286768137894,
+      "loss": 0.0567,
+      "step": 61000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018031800152265137,
+      "loss": 0.0561,
+      "step": 62000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00017701790416972107,
+      "loss": 0.0558,
+      "step": 63000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017370423352948002,
+      "loss": 0.0557,
+      "step": 64000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017037865432778442,
+      "loss": 0.0552,
+      "step": 65000
+    },
+    {
+      "epoch": 4.6,
+      "eval_cer": 0.10578063347635784,
+      "eval_loss": 0.06745224446058273,
+      "eval_runtime": 79.5425,
+      "eval_samples_per_second": 62.859,
+      "eval_steps_per_second": 0.126,
+      "eval_wer": 0.2978,
+      "step": 65000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00016704283727312792,
+      "loss": 0.0546,
+      "step": 66000
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00016369845821730933,
+      "loss": 0.0544,
+      "step": 67000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016034719731351576,
+      "loss": 0.0541,
+      "step": 68000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00015699073817224542,
+      "loss": 0.0537,
+      "step": 69000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00015363076701549341,
+      "loss": 0.0531,
+      "step": 70000
+    },
+    {
+      "epoch": 4.96,
+      "eval_cer": 0.10444382155717027,
+      "eval_loss": 0.06512928009033203,
+      "eval_runtime": 79.8169,
+      "eval_samples_per_second": 62.643,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.2962,
+      "step": 70000
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00015026897182962535,
+      "loss": 0.0525,
+      "step": 71000
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00014690704151736484,
+      "loss": 0.0509,
+      "step": 72000
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0001435466650493204,
+      "loss": 0.0509,
+      "step": 73000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0001401895306154785,
+      "loss": 0.0504,
+      "step": 74000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0001368373247770885,
+      "loss": 0.0504,
+      "step": 75000
+    },
+    {
+      "epoch": 5.31,
+      "eval_cer": 0.10291295629487483,
+      "eval_loss": 0.06379809230566025,
+      "eval_runtime": 79.7333,
+      "eval_samples_per_second": 62.709,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.2864,
+      "step": 75000
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00013349173161936601,
+      "loss": 0.0502,
+      "step": 76000
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00013015443190544018,
+      "loss": 0.0495,
+      "step": 77000
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00012682710223196978,
+      "loss": 0.0494,
+      "step": 78000
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00012351141418685273,
+      "loss": 0.0496,
+      "step": 79000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00012020903350945151,
+      "loss": 0.0492,
+      "step": 80000
+    },
+    {
+      "epoch": 5.67,
+      "eval_cer": 0.10310700963798271,
+      "eval_loss": 0.06157052516937256,
+      "eval_runtime": 79.8007,
+      "eval_samples_per_second": 62.656,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.2888,
+      "step": 80000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00011692161925375738,
+      "loss": 0.0485,
+      "step": 81000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00011365082295491233,
+      "loss": 0.0484,
+      "step": 82000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00011039828779950928,
+      "loss": 0.0481,
+      "step": 83000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0001071656478000858,
+      "loss": 0.048,
+      "step": 84000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00010395452697422703,
+      "loss": 0.0474,
+      "step": 85000
+    },
+    {
+      "epoch": 6.02,
+      "eval_cer": 0.10228767330041613,
+      "eval_loss": 0.06011258438229561,
+      "eval_runtime": 80.0834,
+      "eval_samples_per_second": 62.435,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.2868,
+      "step": 85000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00010076653852868962,
+      "loss": 0.0459,
+      "step": 86000
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 9.760328404895728e-05,
+      "loss": 0.0457,
+      "step": 87000
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 9.446635269463389e-05,
+      "loss": 0.0453,
+      "step": 88000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 9.135732040107998e-05,
+      "loss": 0.0456,
+      "step": 89000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 8.82777490876921e-05,
+      "loss": 0.0453,
+      "step": 90000
+    },
+    {
+      "epoch": 6.38,
+      "eval_cer": 0.1022661118178486,
+      "eval_loss": 0.05900218337774277,
+      "eval_runtime": 79.8035,
+      "eval_samples_per_second": 62.654,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.284,
+      "step": 90000
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 8.522918587322418e-05,
+      "loss": 0.0452,
+      "step": 91000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 8.22131622985438e-05,
+      "loss": 0.0451,
+      "step": 92000
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.923119355721526e-05,
+      "loss": 0.045,
+      "step": 93000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.628477773429438e-05,
+      "loss": 0.0448,
+      "step": 94000
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.3375395053719e-05,
+      "loss": 0.0446,
+      "step": 95000
+    },
+    {
+      "epoch": 6.73,
+      "eval_cer": 0.10045494728217512,
+      "eval_loss": 0.05785335227847099,
+      "eval_runtime": 79.8939,
+      "eval_samples_per_second": 62.583,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.2766,
+      "step": 95000
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 7.050450713467216e-05,
+      "loss": 0.0439,
+      "step": 96000
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.767355625729215e-05,
+      "loss": 0.0439,
+      "step": 97000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.488396463809805e-05,
+      "loss": 0.0438,
+      "step": 98000
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.213713371549497e-05,
+      "loss": 0.0434,
+      "step": 99000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.943444344571755e-05,
+      "loss": 0.042,
+      "step": 100000
+    },
+    {
+      "epoch": 7.08,
+      "eval_cer": 0.09978654132258134,
+      "eval_loss": 0.05739114433526993,
+      "eval_runtime": 79.8794,
+      "eval_samples_per_second": 62.594,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.277,
+      "step": 100000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.6777251609566e-05,
+      "loss": 0.042,
+      "step": 101000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.416689313028253e-05,
+      "loss": 0.0421,
+      "step": 102000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.1604679402911015e-05,
+      "loss": 0.0419,
+      "step": 103000
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.9091897635476726e-05,
+      "loss": 0.0418,
+      "step": 104000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.6629810202317166e-05,
+      "loss": 0.0417,
+      "step": 105000
+    },
+    {
+      "epoch": 7.44,
+      "eval_cer": 0.09941999611893314,
+      "eval_loss": 0.056080542504787445,
+      "eval_runtime": 79.81,
+      "eval_samples_per_second": 62.649,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.2742,
+      "step": 105000
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.421965400988918e-05,
+      "loss": 0.0416,
+      "step": 106000
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.1862639875370016e-05,
+      "loss": 0.0417,
+      "step": 107000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.955995191836581e-05,
+      "loss": 0.0414,
+      "step": 108000
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.731274696603201e-05,
+      "loss": 0.0414,
+      "step": 109000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.512215397190527e-05,
+      "loss": 0.0411,
+      "step": 110000
+    },
+    {
+      "epoch": 7.79,
+      "eval_cer": 0.1000237176308243,
+      "eval_loss": 0.0557967908680439,
+      "eval_runtime": 79.8362,
+      "eval_samples_per_second": 62.628,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.2756,
+      "step": 110000
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.298927344873821e-05,
+      "loss": 0.0407,
+      "step": 111000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.091517691562284e-05,
+      "loss": 0.0407,
+      "step": 112000
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.890090635967912e-05,
+      "loss": 0.0412,
+      "step": 113000
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.694747371258042e-05,
+      "loss": 0.0398,
+      "step": 114000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.505586034217802e-05,
+      "loss": 0.0394,
+      "step": 115000
+    },
+    {
+      "epoch": 8.15,
+      "eval_cer": 0.09991591021798658,
+      "eval_loss": 0.05538313463330269,
+      "eval_runtime": 79.9108,
+      "eval_samples_per_second": 62.57,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.2758,
+      "step": 115000
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.3227016559480356e-05,
+      "loss": 0.0396,
+      "step": 116000
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.1461861141234628e-05,
+      "loss": 0.0396,
+      "step": 117000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.9761280868350654e-05,
+      "loss": 0.0397,
+      "step": 118000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.8126130080398725e-05,
+      "loss": 0.0395,
+      "step": 119000
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.655723024640553e-05,
+      "loss": 0.0394,
+      "step": 120000
+    },
+    {
+      "epoch": 8.5,
+      "eval_cer": 0.09993747170055413,
+      "eval_loss": 0.05500805750489235,
+      "eval_runtime": 79.8857,
+      "eval_samples_per_second": 62.589,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.275,
+      "step": 120000
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.5055369552163509e-05,
+      "loss": 0.0394,
+      "step": 121000
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.3621302504261194e-05,
+      "loss": 0.0395,
+      "step": 122000
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.2255749551033205e-05,
+      "loss": 0.0393,
+      "step": 123000
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.0959396720620555e-05,
+      "loss": 0.0393,
+      "step": 124000
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 9.732895276323099e-06,
+      "loss": 0.0392,
+      "step": 125000
+    },
+    {
+      "epoch": 8.86,
+      "eval_cer": 0.09957092649690592,
+      "eval_loss": 0.054800018668174744,
+      "eval_runtime": 80.1483,
+      "eval_samples_per_second": 62.384,
+      "eval_steps_per_second": 0.125,
+      "eval_wer": 0.2756,
+      "step": 125000
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 8.576861389416939e-06,
+      "loss": 0.0391,
+      "step": 126000
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.4918758296016734e-06,
+      "loss": 0.039,
+      "step": 127000
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 6.478483673232537e-06,
+      "loss": 0.0388,
+      "step": 128000
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.537194029484349e-06,
+      "loss": 0.0383,
+      "step": 129000
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.668479784584583e-06,
+      "loss": 0.0383,
+      "step": 130000
+    },
+    {
+      "epoch": 9.21,
+      "eval_cer": 0.09931218870609543,
+      "eval_loss": 0.054751720279455185,
+      "eval_runtime": 80.4739,
+      "eval_samples_per_second": 62.132,
+      "eval_steps_per_second": 0.124,
+      "eval_wer": 0.2748,
+      "step": 130000
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.872777364244245e-06,
+      "loss": 0.0387,
+      "step": 131000
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.150486514405798e-06,
+      "loss": 0.0385,
+      "step": 132000
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.50197010041826e-06,
+      "loss": 0.0385,
+      "step": 133000
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.9275539247405437e-06,
+      "loss": 0.0384,
+      "step": 134000
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4275265632643806e-06,
+      "loss": 0.0386,
+      "step": 135000
+    },
+    {
+      "epoch": 9.56,
+      "eval_cer": 0.09924750425839281,
+      "eval_loss": 0.05460319668054581,
+      "eval_runtime": 80.404,
+      "eval_samples_per_second": 62.186,
+      "eval_steps_per_second": 0.124,
+      "eval_wer": 0.2752,
+      "step": 135000
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.0021392203391932e-06,
+      "loss": 0.0383,
+      "step": 136000
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 6.516056025718585e-07,
+      "loss": 0.0383,
+      "step": 137000
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.76101811464391e-07,
+      "loss": 0.0383,
+      "step": 138000
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.7576625494392648e-07,
+      "loss": 0.0384,
+      "step": 139000
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.0699577829210744e-08,
+      "loss": 0.0387,
+      "step": 140000
+    },
+    {
+      "epoch": 9.92,
+      "eval_cer": 0.09937687315379806,
+      "eval_loss": 0.05459373816847801,
+      "eval_runtime": 80.3513,
+      "eval_samples_per_second": 62.227,
+      "eval_steps_per_second": 0.124,
+      "eval_wer": 0.2752,
+      "step": 140000
+    }
+  ],
+  "max_steps": 141160,
+  "num_train_epochs": 10,
+  "total_flos": 3.547844813555835e+17,
+  "trial_name": null,
+  "trial_params": null
+}

byT5/Charsiu/g2p_multilingual_byT5_tiny_8_layers/.gitattributes

@@ -0,0 +1,27 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

byT5/Charsiu/g2p_multilingual_byT5_tiny_8_layers/config.json

@@ -0,0 +1,29 @@
+{
+  "_name_or_path": "/home/patrick/t5/byt5-small",
+  "architectures": [
+    "T5ForConditionalGeneration"
+  ],
+  "d_ff": 512,
+  "d_kv": 64,
+  "d_model": 256,
+  "decoder_start_token_id": 0,
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-gelu",
+  "gradient_checkpointing": false,
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "num_decoder_layers": 2,
+  "num_heads": 6,
+  "num_layers": 6,
+  "pad_token_id": 0,
+  "relative_attention_num_buckets": 32,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "ByT5Tokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.16.2",
+  "use_cache": true,
+  "vocab_size": 384
+}

byT5/Charsiu/g2p_multilingual_byT5_tiny_8_layers/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa6384833adad3313930a29e2fd80c8366fe1132a1a314cee15f5ca1f0b81b06
+size 29159133

byT5/Charsiu/g2p_multilingual_byT5_tiny_8_layers/source.txt

@@ -0,0 +1 @@
+https://huggingface.co/charsiu/g2p_multilingual_byT5_tiny_8_layers