[2025-10-10 13:10:41,462] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:24741] baseline 0.000GB (+0.000GB allocated, +0.002GB reserved) [2025-10-10 13:10:41,462] [INFO] [axolotl.cli.config.load_cfg:248] [PID:24741] config: { "activation_offloading": false, "adapter": "qlora", "axolotl_config_path": "config.yaml", "base_model": "nferruz/ProtGPT2", "base_model_config": "nferruz/ProtGPT2", "batch_size": 2, "bf16": false, "capabilities": { "bf16": true, "compute_capability": "sm_75", "fp8": false, "n_gpu": 1, "n_node": 1 }, "context_parallel_size": 1, "dataloader_num_workers": 1, "dataloader_pin_memory": true, "dataloader_prefetch_factor": 256, "dataset_processes": 2, "datasets": [ { "ds_type": "json", "message_property_mappings": { "content": "content", "role": "role" }, "path": "/content/sequences_tokenized.jsonl", "trust_remote_code": false } ], "ddp": false, "device": "cuda:0", "dion_rank_fraction": 1.0, "dion_rank_multiple_of": 1, "env_capabilities": { "torch_version": "2.8.0" }, "eval_batch_size": 2, "eval_causal_lm_metrics": [ "sacrebleu", "comet", "ter", "chrf" ], "eval_max_new_tokens": 128, "eval_sample_packing": true, "eval_steps": 0.01, "eval_table_size": 0, "experimental_skip_move_to_device": true, "fp16": true, "gradient_accumulation_steps": 1, "gradient_checkpointing": true, "gradient_checkpointing_kwargs": { "use_reentrant": true }, "group_by_length": false, "hub_model_id": "ProtGPT2-Oxido", "include_tkps": true, "is_falcon_derived_model": false, "is_llama_derived_model": false, "is_mistral_derived_model": false, "learning_rate": 0.002, "lisa_layers_attribute": "model.layers", "load_best_model_at_end": false, "load_in_4bit": true, "load_in_8bit": false, "local_rank": 0, "logging_steps": 100, "lora_alpha": 16, "lora_dropout": 0.05, "lora_r": 32, "lora_target_linear": true, "loraplus_lr_embedding": 1e-06, "lr_scheduler": "cosine", "mean_resizing_embeddings": false, "micro_batch_size": 2, "model_config_type": "gpt2", "num_epochs": 3.0, "optimizer": "paged_adamw_32bit", "output_dir": "./qlora-out", "pad_to_sequence_len": true, "pretrain_multipack_attn": true, "profiler_steps_start": 0, "qlora_sharded_model_loading": false, "ray_num_workers": 1, "resources_per_worker": { "GPU": 1 }, "sample_packing": true, "sample_packing_bin_size": 200, "sample_packing_group_size": 100000, "save_only_model": false, "save_safetensors": true, "save_strategy": "epoch", "sequence_len": 2048, "shuffle_before_merging_datasets": false, "shuffle_merged_datasets": true, "skip_prepare_dataset": false, "special_tokens": { "eos_token": "<|endoftext|>", "pad_token": "<|endoftext|>" }, "streaming_multipack_buffer_size": 10000, "strict": false, "tensor_parallel_size": 1, "tf32": false, "tiled_mlp_use_original_mlp": true, "tokenizer_config": "nferruz/ProtGPT2", "tokenizer_save_jinja_files": true, "tokenizer_type": "AutoTokenizer", "torch_dtype": "torch.float16", "train_on_inputs": false, "trl": { "log_completions": false, "mask_truncated_completions": false, "ref_model_mixup_alpha": 0.9, "ref_model_sync_steps": 64, "scale_rewards": true, "sync_ref_model": false, "use_vllm": false, "vllm_server_host": "0.0.0.0", "vllm_server_port": 8000 }, "type_of_model": "AutoModelForCausalLM", "use_ray": false, "val_set_size": 0.02, "vllm": { "device": "auto", "dtype": "auto", "gpu_memory_utilization": 0.9, "host": "0.0.0.0", "port": 8000 }, "warmup_steps": 100, "weight_decay": 0.0, "world_size": 1 } [2025-10-10 13:10:42,465] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:278] [PID:24741] EOS: 0 / <|endoftext|> [2025-10-10 13:10:42,465] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:279] [PID:24741] BOS: 0 / <|endoftext|> [2025-10-10 13:10:42,465] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:280] [PID:24741] PAD: 0 / <|endoftext|> [2025-10-10 13:10:42,465] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:24741] UNK: 0 / <|endoftext|> [2025-10-10 13:10:42,465] [INFO] [axolotl.loaders.tokenizer.load_tokenizer:295] [PID:24741] No Chat template selected. Consider adding a chat template for easier inference. [2025-10-10 13:10:42,466] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:476] [PID:24741] Unable to find prepared dataset in last_run_prepared/120d8e2ed44f3c537dc9a20773f86561 [2025-10-10 13:10:42,466] [INFO] [axolotl.utils.data.sft._load_raw_datasets:320] [PID:24741] Loading raw datasets... [2025-10-10 13:10:42,466] [WARNING] [axolotl.utils.data.sft._load_raw_datasets:322] [PID:24741] Processing datasets during training can lead to VRAM instability. Please pre-process your dataset using `axolotl preprocess path/to/config.yml`. [2025-10-10 13:10:42,802] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:24741] Loading dataset: /content/sequences_tokenized.jsonl with base_type: None and prompt_style: None [2025-10-10 13:10:42,820] [INFO] [axolotl.utils.data.utils.handle_long_seq_in_dataset:218] [PID:24741] min_input_len: 6 [2025-10-10 13:10:42,821] [INFO] [axolotl.utils.data.utils.handle_long_seq_in_dataset:220] [PID:24741] max_input_len: 512 Dropping Long Sequences (>2048) (num_proc=2): 0% 0/6304 [00:002048) (num_proc=2): 16% 1000/6304 [00:00<00:01, 2728.50 examples/s] Dropping Long Sequences (>2048) (num_proc=2): 48% 3000/6304 [00:00<00:00, 5756.81 examples/s] Dropping Long Sequences (>2048) (num_proc=2): 79% 5000/6304 [00:00<00:00, 7979.32 examples/s] Dropping Long Sequences (>2048) (num_proc=2): 100% 6304/6304 [00:00<00:00, 6859.52 examples/s] Drop Samples with Zero Trainable Tokens (num_proc=2): 0% 0/6304 [00:00 [2025-10-10 13:10:57,214] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:279] [PID:24741] BOS: 0 / <|endoftext|> [2025-10-10 13:10:57,214] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:280] [PID:24741] PAD: 0 / <|endoftext|> [2025-10-10 13:10:57,214] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:24741] UNK: 0 / <|endoftext|> [2025-10-10 13:10:57,214] [INFO] [axolotl.loaders.tokenizer.load_tokenizer:295] [PID:24741] No Chat template selected. Consider adding a chat template for easier inference. [2025-10-10 13:10:57,215] [DEBUG] [axolotl.train.setup_model_and_tokenizer:74] [PID:24741] Loading model [2025-10-10 13:10:57,333] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:87] [PID:24741] Patched Trainer.evaluation_loop with nanmean loss calculation [2025-10-10 13:10:57,334] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:138] [PID:24741] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation [2025-10-10 13:10:57,335] [INFO] [axolotl.loaders.patch_manager._apply_multipack_patches:301] [PID:24741] Applying multipack dataloader patch for sample packing... [2025-10-10 13:11:27,461] [WARNING] [axolotl.loaders.model._adjust_model_config:273] [PID:24741] increasing model.config.max_position_embeddings from 1024 to 2048 [2025-10-10 13:11:27,467] [INFO] [axolotl.loaders.model._prepare_model_for_quantization:863] [PID:24741] converting PEFT model w/ prepare_model_for_kbit_training [2025-10-10 13:11:27,479] [INFO] [axolotl.loaders.model._configure_embedding_dtypes:345] [PID:24741] Converting modules to torch.float16 [2025-10-10 13:11:27,481] [DEBUG] [axolotl.loaders.model.log_gpu_memory_usage:127] [PID:24741] Memory usage after model load 0.849GB (+0.849GB allocated, +0.918GB reserved) [2025-10-10 13:11:27,482] [INFO] [axolotl.loaders.adapter.load_lora:80] [PID:24741] found linear modules: ['c_attn', 'c_fc', 'c_proj'] trainable params: 23,592,960 || all params: 797,623,040 || trainable%: 2.9579 [2025-10-10 13:11:27,888] [DEBUG] [axolotl.loaders.model.log_gpu_memory_usage:127] [PID:24741] after adapters 0.618GB (+0.618GB allocated, +1.012GB reserved) [2025-10-10 13:11:39,738] [INFO] [axolotl.train.save_initial_configs:398] [PID:24741] Pre-saving adapter config to ./qlora-out... [2025-10-10 13:11:39,738] [INFO] [axolotl.train.save_initial_configs:402] [PID:24741] Pre-saving tokenizer to ./qlora-out... [2025-10-10 13:11:39,828] [INFO] [axolotl.train.save_initial_configs:407] [PID:24741] Pre-saving model config to ./qlora-out... [2025-10-10 13:11:39,836] [INFO] [axolotl.train.execute_training:196] [PID:24741] Starting trainer... [2025-10-10 13:11:45,415] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:24741] generate_batches time: 1.8461699485778809 [2025-10-10 13:11:47,771] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:24741] generate_batches time: 2.3554582595825195 [2025-10-10 13:11:49,329] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:24741] generate_batches time: 1.5573019981384277 [2025-10-10 13:11:51,006] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:24741] generate_batches time: 1.6764421463012695 [2025-10-10 13:11:51,006] [INFO] [axolotl.utils.samplers.multipack.calc_min_len:438] [PID:24741] gather_len_batches: [183] 0% 0/549 [00:00