-
fix GQA compat PyTorch 2.4 and strip _orig_mod. on checkpoint load
Juan S Santillana
-
Upload config_600m.json with huggingface_hub
-
Upload training_v2/data/sft_dataset.py with huggingface_hub
-
Upload training_v2/data/curriculum_dataset.py with huggingface_hub
-
Upload training_v2/train/utils.py with huggingface_hub
-
Upload training_v2/train/pretrain_750m.py with huggingface_hub
-
Upload training_v2/model/transformer_750m.py with huggingface_hub
-
initial commit