Fix for RuntimeError: FlashAttention only support fp16 and bf16 data type during fine tuning.

Resolve - 196 [rank0]: triton.runtime.autotuner.OutOfResources: out of resource: shared memory, Required: 180224, Hardware limit: 101376. Reducing block sizes or `num_stages` may help.

Move flash_attn assert from __init__ into calling func (#32)

Update README.md

Update tokenization_phi3_small.py (#14)

Add link to Phi-3 vision ONNX models

Update README.md

Add attention_bias to make TGI work (#4)

Update README

chore(root): Initial files upload.

Initial commit