Text Generation
Transformers
PyTorch
English
gpt2
feature-extraction
causal-lm
text-generation-inference

How to merge the 2 bin files into the pytorch_model.bin file for usage?

#2
by spanielrassler - opened

I'm sure this is a stupid questions, but how do I merge the bin files into one file that can be used as a model? Final model name should be pytorch_model.bin but with 2 separate .bin files not sure how to do that.

The transformers library automatically shards large models with the save_pretrained() method into partial checkpoints and an index that maps parameter names to the files they are stored in. The model be loaded by passing the model directory (e.g., ./my_model_directory/) to from_pretrained().

These links may be helpful https://huggingface.co/docs/transformers/big_models and https://github.com/huggingface/transformers/blob/v4.27.2/src/transformers/modeling_utils.py#L1833

Sign up or log in to comment