logikon-ai / cot-eval

A framework for evaluating the effectiveness of chain-of-thought reasoning in language models.
https://huggingface.co/spaces/logikon/open_cot_leaderboard
MIT License
5 stars 1 forks source link

Evaluate: jetmoe/jetmoe-8b | -8b-sft | -8b-chat #47

Open ggbetz opened 3 months ago

ggbetz commented 3 months ago

Check upon issue creation:

Parameters:

NEXT_MODEL_PATH=jetmoe/jetmoe-8b
NEXT_MODEL_REVISION=main
NEXT_MODEL_PRECISION=bfloat16
MAX_LENGTH=2048 
GPU_MEMORY_UTILIZATION=0.5
VLLM_SWAP_SPACE=8

ToDos:

yakazimir commented 2 months ago

I think our transformers version might be out of date: see here: https://github.com/huggingface/text-generation-inference/issues/1620


2024-05-13T17:16:46.166834523Z 2024-05-13 17:16:46,166 - root - INFO - Loading vLLM model jetmoe/jetmoe-8b
2024-05-13T17:16:48.518010161Z Traceback (most recent call last):
2024-05-13T17:16:48.518059680Z   File "/usr/local/bin/cot-eval", line 8, in <module>
2024-05-13T17:16:48.518069710Z     sys.exit(main())
2024-05-13T17:16:48.518076080Z   File "/workspace/cot-eval/src/cot_eval/__main__.py", line 149, in main
2024-05-13T17:16:48.518202040Z     llm = VLLM(
2024-05-13T17:16:48.518229400Z   File "/usr/local/lib/python3.10/dist-packages/langchain_core/load/serializable.py", line 120, in __init__
2024-05-13T17:16:48.518319969Z     super().__init__(**kwargs)
2024-05-13T17:16:48.518333769Z   File "/usr/local/lib/python3.10/dist-packages/pydantic/v1/main.py", line 341, in __init__
2024-05-13T17:16:48.518495708Z     raise validation_error
2024-05-13T17:16:48.518561108Z pydantic.v1.error_wrappers.ValidationError: 1 validation error for VLLM
2024-05-13T17:16:48.518565458Z __root__
2024-05-13T17:16:48.518568898Z   The checkpoint you are trying to load has model type `jetmoe` but Transformers does not recognize this architecture. This could be because of an issue with the checkpoint, or because your version of Transformers is out of date. (type=value_error)``` 
ggbetz commented 2 months ago

https://github.com/logikon-ai/cot-eval/issues/54

and:

https://github.com/vllm-project/vllm/issues/3995