expand seq_len to 1024

README.md

@@ -11,13 +11,13 @@ pipeline_tag: text-generation
 tags:
 - chat
 widget:
-- text: "<Human>: Hello <eoh> <Assistant>:"
+- text: "<Human>:Hello<Assistant>:"
   example_title: "Hello"
-- text: "<Human>: 你好 <eoh> <Assistant>:"
+- text: "<Human>:你好<Assistant>:"
   example_title: "你好"
-- text: "<Human>: What should I do if I can't sleep at night? <eoh> <Assistant>:"
+- text: "<Human>:What should I do if I can't sleep at night?<Assistant>:"
   example_title: "insomnia"
-- text: "<Human>: 晚上睡不着应该怎么办？ <eoh> <Assistant>:"
+- text: "<Human>:晚上睡不着应该怎么办？<Assistant>:"
   example_title: "失眠"
 inference:
   parameters:
@@ -42,7 +42,7 @@ tokenizer.pad_token_id = tokenizer.eos_token_id
 
 model = AutoModelForCausalLM.from_pretrained('nicholascao/chatbloom-1b7-sft').half()
 
-inputs = tokenizer('<Human>: Hello <eoh> <Assistant>:', return_tensors='pt').to(torch.cuda.current_device())
+inputs = tokenizer('<Human>:Hello<Assistant>:', return_tensors='pt').to(torch.cuda.current_device())
 model.to(torch.cuda.current_device())
 
 output = model.generate(**inputs, max_length=768, do_sample=True, temperature=0.8, top_k=50, early_stopping=True, repetition_penalty=1.05)

config.json

@@ -21,7 +21,7 @@
   "offset_alibi": 100,
   "pad_token_id": 3,
   "pretraining_tp": 2,
-  "seq_length": 4096,
+  "seq_length": 1024,
   "skip_bias_add": true,
   "skip_bias_add_qkv": false,
   "slow_but_exact": false,

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:076ce6313d3e4895c711440cf34ee5205223b9bd26286fa2fa41e9ce2f92579d
-size 3444055875
+oid sha256:46eebad455425439b1dbf242f96c5d19317c0d4e14768a565ca5ca8d63a00ce6
+size 3444047683