FastChat
FastChat copied to clipboard
lm-sys
An open platform for training, serving, and evaluating large language models. Release repo for Vicuna and Chatbot Arena.
When running ``` python ../fastchat/eval/get_model_answer.py --model-id 1 \ --model-path ../vicuna \ --question-file t.jsonl \ --answer-file answer.jsonl \ --num-gpus 2 ``` I get the following error: ``` 2023-05-05 17:23:57,395 INFO worker.py:1616...
Is it planned to also release the instruction & responses of each model, not just the winner model ?
Currently we know from the dataset which model is better than another model. It would be useful to also have the different responses and their respective inputs, but I don't...
Hi, since I updated fastchat to version 0.2.2 I can no longer make the 4-bit GPTQ work because I get this error: `python3 -m fastchat.serve.cli --model-path models/TheBloke_vicuna-7B-1.1-GPTQ-4bit-128g --wbits 4 --groupsize...
Hi,, I have been trying to reproduce vicuna in 4xL4 which is 4*24=96 gb vram. What approach should I follow? Train.py oom error Train_mem.py oom error Left is Lora but...
https://huggingface.co/mosaicml/mpt-7b-storywriter The model appears to be a great alternative to LLaMA from surface observations, particularly for commercial use. I expect many groups to build upon this foundation going forward if...
Hello, I've noticed that in the `FastChat/fastchat/conversation.py` file has code that interfaces with multiple LLMs. I think that as we continue to add more LLMs, there will come a time...
Hi I noticed that Nvidia released the following model https://huggingface.co/nvidia/GPT-2B-001 Is this something we can run on the FastChat platform?
Hi, Can I use GPTQ quantized model to do the inference? ``` https://github.com/qwopqwop200/GPTQ-for-LLaMa ``` Thank you!
hello, i can't load the model vicuna-13B-1.1-GPTQ-4bit-128g . i have installed git clone https://github.com/oobabooga/GPTQ-for-LLaMa.git -b cuda and python 3.9 thanks for your help. **python -m fastchat.serve.cli --model-name TheBloke/vicuna-13B-1.1-GPTQ-4bit-128g --wbits 4...
