We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Following the GPT Pretraining section in the Megatron-LM repo, we are able to successfully train a model using Megatron-LM
Megatron-LM
I saw pointers on how to convert from HF to nemo. For example, this conversion script convert_llama_hf_to_nemo.py
nemo
However I did not see any examples of converting a ckpt saved using the Megatron-LM to nemo format . Are there any examples for this?
p.s. I am thinking of doing this conversion to nemo, so I can use tools like Nemo-Aligner for post-training
Nemo-Aligner
The text was updated successfully, but these errors were encountered:
There is a script for that: <NeMo_ROOT_FOLDER>/examples/nlp/language_modeling/megatron_lm_ckpt_to_nemo.py
<NeMo_ROOT_FOLDER>/examples/nlp/language_modeling/megatron_lm_ckpt_to_nemo.py
Look up this guide: https://docs.nvidia.com/nemo-framework/user-guide/latest/nemotoolkit/checkpoints/convert_mlm.html
Also this issue I opened might be relevant to you: #10480
Sorry, something went wrong.
thanks @aimarz . let me try it out on the Megatron-LM ckpts we pre-trained
No branches or pull requests
Following the GPT Pretraining section in the
Megatron-LM
repo, we are able to successfully train a model usingMegatron-LM
I saw pointers on how to convert from HF to
nemo
. For example, this conversion script convert_llama_hf_to_nemo.pyHowever I did not see any examples of converting a ckpt saved using the
Megatron-LM
tonemo
format . Are there any examples for this?p.s. I am thinking of doing this conversion to
nemo
, so I can use tools likeNemo-Aligner
for post-trainingThe text was updated successfully, but these errors were encountered: