Chuckle

[Feature] large data storage in Ray object store

3

### Search before asking - [X] I had searched in the [issues](https://github.com/ray-project/ray/issues) and found no similar feature requirement. ### Description Hi, I have utilized ray tune to do the knowledge...

loss 0 with load_in_8bit = True

7

Hello, I came across an issue regarding loss remaining at 0 when using PEFT for the chatglm model. I found that the model works well with fp16. For your reference,...

### Is your feature request related to a problem? Please describe. _No response_ ### Solutions hello, 对于微调的结果是否可以提供一个vicuna类似的评估结果，我感觉比较难确定我们在单一垂直领域任务下继续微调出来的效果可以打败你们原生大数据多任务微调的结果 ### Additional context _No response_

Multiple GPUs not work

1

it still loads the model only in one gpu and gives me OOM. I have exactly followed the instruction in the readme.

**Describe the bug** Hi there, I have set torch.no_grad() and torch.cuda.empty_cache(), but the GPU still encounters out-of-memory (OOM) errors after a few inferences. My torch version is 1.13.1, deepspeed version...

hello，是否可以提供一份完整的不同任务下的中文任务prompt样例，还是基于英文的翻译一下就可以了

实验结果

3

my friend，需要一个实验分析和结果来证明你的实验的有效性