Open-Llama icon indicating copy to clipboard operation
Open-Llama copied to clipboard

Published 20 hours ago • verified publisher icon s-JoL
→

Metadata

The complete training code of the open-source high-performance Llama model, including the full process from pre-training to RLHF.

Results 16 Open-Llama issues
Sort by recently updated
recently updated
newest added

触发了重复生成的问题

这种重复生成的问题是什么原因导致的，好像每次当inputs长度比较长的时候就容易触发重复生成，这个问题可以怎么解决？

Expert68 avatar Expert68

training benchmarking question: unable to reproduce the training speed

Hi @s-JoL, thank you very much for your efforts making llama open to the community. I got some question when benchmarking the training speed: My environment: One node with 8...

tong-zeng avatar tong-zeng

v2的模型使用了RLHF吗

v2的模型使用了RLHF吗，为什么效果提升了这么多

Expert68 avatar Expert68

Can not send text

![image](https://github.com/s-JoL/Open-Llama/assets/21303438/8d7a30b8-be46-4a0a-a4b9-abf6af702d85)

lucasjinreal avatar lucasjinreal

关于多机多卡训练

您好，我在多机多卡训练启动时遇到了一些报错，请问您多机训练时ds_stage3_offload中的参数具体是怎么设置的？

lemonsis avatar lemonsis

Is multi-node (multiple machines) training with zero3 tested?

thanks

lipiji avatar lipiji

← Metadata

863
Stars
63
Forks
26
Watchers

Owner

verified publisher icon s-JoL

Metadata

The complete training code of the open-source high-performance Llama model, including the full process from pre-training to RLHF.

Back