flybird11111 comments

Results 68 comments of


                                            flybird11111

when I uese hybrid_parallel, and set the enable_fused_normalization = True. I can't run the code, here are some error: RuntimeError: Failed to replace input_layernorm of type LlamaRMSNorm with FusedRMSNorm with the exception: Please install apex from source (https://github.com/NVIDIA/apex) to use the fused RMS normalization kernel. Please check your model configuration or sharding policy, you can set up an issue for us to help you as well. However, I have install the apex, it will still occur. How can i solve it?

Can you share your pip list and your cuda version?

[BUG]: On the eight-card A100, testing the 'examples/language/llama2' with the 'gemini_auto' plugin resulted in an 'out of memory' error."

Hi, how about trying to set offload_optim_frac and offload_param_frac to 1.0?

[BUG]: ImportError: cannot import name 'MOE_CONTEXT' from partially initialized module

Hi, 'colossalai.nn.layer.moe' will be deprecated soon. you can follow the open-moe, https://github.com/hpcaitech/ColossalAI/blob/main/applications/ColossalMoE/train.py

代码bugs太多了[BUG]:

Could you please provide more details about the errors?

代码bugs太多了[BUG]:

Thank you. We will address these issues.

> Besides, in your readme, you tell that I need to use xformers0.0.12. However, in the code > > https://github.com/hpcaitech/ColossalAI/blob/d4fb7bfda7a2da5480e1187e8d3e40884b42ba11/applications/Chat/coati/kernels/opt_attn.py#L72 > > You use attn_bias which is not found in...

[BUG]: The accuracy of vit is very low

Colossal AI is not the primary cause of this result.

build docker failed

> I just use Docker with MacOS, not build on MacOS system. Does it matter? If you are using Docker on macOS to run your training environment but not building...

[DOC]: 公式乱码

已修复

[BUG]: How can run examples/images/diffusion with use_ema

not yet