what's the difference between this repo with "santacoder-megatron"?
#4
by
lsc4452372
- opened
Hi, I noticed that you guys also have a repo call "santacoder-megatron" (https://huggingface.co/bigcode/santacoder-megatron), where the files are more similar to other megatron-based repos. Therefore, I wonder to know what's the difference between this repo ("gpt_bigcode-santacoder") with "santacoder-megatron"?
This is a transformers version of the model. The megatron version can't be loaded with transformers.
Edit: confusion comes from the readme, sorry we'll update it
This is a transformers version of the model. The megatron version can't be loaded with transformers.
Edit: confusion comes from the readme, sorry we'll update it
Thanks for your reply~
christopher
changed discussion status to
closed