Add batched generation support to model/ package

Fix CPU inference; add KV cache, FIM example, HumanEval results

Upload train.log with huggingface_hub

Upload tokenizer_src/tokenizer_utils.py with huggingface_hub

Upload tokenizer/tokenizer.json with huggingface_hub

Upload sft_train.log with huggingface_hub

Upload requirements.txt with huggingface_hub

Upload README.md with huggingface_hub

Upload model.safetensors with huggingface_hub

Upload model/transformer.py with huggingface_hub

Upload model/pycraft_model.py with huggingface_hub

Upload model/feedforward.py with huggingface_hub

Upload model/config.py with huggingface_hub

Upload model/attention.py with huggingface_hub

Upload model/__init__.py with huggingface_hub

Upload config.json with huggingface_hub

initial commit