Matthias Reso

Maybe as addition context: I am trying to set the map_location during loading because I want to combine tokenizer and model with a Sequential transformation to save them into the...

Thanks @parmeet, that makes sense. From a user perspective I would prefer a more elaborate error message. Same when loading the tokenizer but torchtext is not installed. In this case...

Just noticed that I get "Aborted (core dumped)" even if torchtext is installed but I do not call "import torchtext" before executing torch.jit.load. Thats a tricky situation as in our...

Closing this draft as the main contributions are already in main after merging https://github.com/pytorch/serve/pull/2984.

Closing this as we'll likely not add more datasets and the [use case of continued full-parameter update is already supported and described](https://github.com/meta-llama/llama-recipes/blob/main/recipes/finetuning/multigpu_finetuning.md#with-fsdp--peft). For massive parallel pre-training have a look at...

Hi, thanks for filing the issue! The behavior is indeed unfortunate in your case and the proposed solution sounds reasonable to me. The only edge case I could come up...

@MuhsinFatih Great! Looking forward to your PR! Let me know if you need any help with it! Agreed on the solution for users using bogus filenames. Would be great to...

Thanks for your contribution! I'll close this PR as the notebook was already updated in [another PR](https://github.com/meta-llama/llama-recipes/pull/458).

Thanks for the suggestions, we welcome contributions from the community.

Hi @waterluck Q1: What looks a bit weird to me is that the __0_X.distcp files get bigger when you you store the optimizer as well. Will need to look into...