Ziang Ye

单机多卡运行时报错 has parameters that were not used in producing loss

6

### 🐛 bug 说明 **使用指令** CUDA_VISIBLE_DEVICES=2,3 accelerate launch --num_processes 2 path_to_train_m3e.py path_to_model path_to_dataset \ --output-dir output_dir **报错信息** RuntimeError: Expected to have finished reduction in the prior iteration before starting a...

Thanks for your great work! I wonder if you can opensource the original dataset that havn't been filterd by reward, it will be of great use for me.

Adopt to tool_calls format of oai doc fixing function call bugs and Support for parallel function call

3

I'm using deepseek api(which is oai compatible), which i found the code fail to return the function call result: e.g.  > UnprocessableEntityError: Failed to deserialize the JSON body into...