diff --git a/deepspeed_config/ds_stage2.json b/deepspeed_config/ds_stage2.json old mode 100644 new mode 100755 diff --git a/deepspeed_config/ds_stage3.json b/deepspeed_config/ds_stage3.json old mode 100644 new mode 100755 diff --git a/documents/doc_tokenizer.docx b/documents/doc_tokenizer.docx old mode 100644 new mode 100755 diff --git a/src_inference/dolly_inference.py b/src_inference/dolly_inference.py old mode 100644 new mode 100755 index 0701c49..fd238b4 --- a/src_inference/dolly_inference.py +++ b/src_inference/dolly_inference.py @@ -1,8 +1,5 @@ from transformers import AutoModelForCausalLM, AutoTokenizer - -model_name = "Qwen/Qwen3-8B" - -# load the tokenizer and the model +model_name = "" tokenizer = AutoTokenizer.from_pretrained(model_name) model = AutoModelForCausalLM.from_pretrained( model_name, @@ -10,7 +7,7 @@ model = AutoModelForCausalLM.from_pretrained( device_map="auto" ) -# prepare the model input + prompt = "Give me a short introduction to large language model." messages = [ {"role": "user", "content": prompt} @@ -19,20 +16,18 @@ text = tokenizer.apply_chat_template( messages, tokenize=False, add_generation_prompt=True, - enable_thinking=True # Switches between thinking and non-thinking modes. Default is True. + enable_thinking=True ) model_inputs = tokenizer([text], return_tensors="pt").to(model.device) -# conduct text completion generated_ids = model.generate( **model_inputs, max_new_tokens=32768 ) output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist() -# parsing thinking content + try: - # rindex finding 151668 () index = len(output_ids) - output_ids[::-1].index(151668) except ValueError: index = 0 diff --git a/src_llm_moe/__init__.py b/src_llm_moe/__init__.py old mode 100644 new mode 100755 diff --git a/src_llm_moe/__pycache__/__init__.cpython-312.pyc b/src_llm_moe/__pycache__/__init__.cpython-312.pyc deleted file mode 100644 index b72ac68..0000000 Binary files a/src_llm_moe/__pycache__/__init__.cpython-312.pyc and /dev/null differ diff --git a/src_llm_moe/__pycache__/configuration_dollyMoE.cpython-312.pyc b/src_llm_moe/__pycache__/configuration_dollyMoE.cpython-312.pyc deleted file mode 100644 index c71d156..0000000 Binary files a/src_llm_moe/__pycache__/configuration_dollyMoE.cpython-312.pyc and /dev/null differ diff --git a/src_llm_moe/__pycache__/modeling_dollyMoE.cpython-312.pyc b/src_llm_moe/__pycache__/modeling_dollyMoE.cpython-312.pyc deleted file mode 100644 index 682d2fb..0000000 Binary files a/src_llm_moe/__pycache__/modeling_dollyMoE.cpython-312.pyc and /dev/null differ diff --git a/src_llm_moe/configuration_dollyMoE.py b/src_llm_moe/configuration_dollyMoE.py old mode 100644 new mode 100755 diff --git a/src_llm_moe/modeling_dollyMoE.py b/src_llm_moe/modeling_dollyMoE.py old mode 100644 new mode 100755 diff --git a/src_pretrain/pretrain_deepspeed.py b/src_pretrain/pretrain_deepspeed.py old mode 100644 new mode 100755 diff --git a/src_pretrain/pretrain_deepspeed_MoE.py b/src_pretrain/pretrain_deepspeed_MoE.py old mode 100644 new mode 100755 diff --git a/src_pretrain/pretrain_transformers.py b/src_pretrain/pretrain_transformers.py old mode 100644 new mode 100755 diff --git a/test/compute_dollyMoE.py b/test/compute_dollyMoE.py old mode 100644 new mode 100755 diff --git a/test/test_tokenizer.py b/test/test_tokenizer.py old mode 100644 new mode 100755 diff --git a/test/test_tokenizer_weights.py b/test/test_tokenizer_weights.py old mode 100644 new mode 100755 diff --git a/utils/dolly_pretrain_deepspeed.sh b/utils/dolly_pretrain_deepspeed.sh old mode 100644 new mode 100755 diff --git a/utils/dolly_pretrain_torch.sh b/utils/dolly_pretrain_torch.sh old mode 100644 new mode 100755 diff --git a/weights_llm/config.json b/weights_llm/config.json old mode 100644 new mode 100755 diff --git a/weights_llm/generation_config.json b/weights_llm/generation_config.json old mode 100644 new mode 100755 diff --git a/weights_llm/merges.txt b/weights_llm/merges.txt old mode 100644 new mode 100755 diff --git a/weights_llm/tokenizer.json b/weights_llm/tokenizer.json old mode 100644 new mode 100755 diff --git a/weights_llm/tokenizer_config.json b/weights_llm/tokenizer_config.json old mode 100644 new mode 100755 diff --git a/weights_llm/vocab.json b/weights_llm/vocab.json old mode 100644 new mode 100755