make sure input_ids do not contain negative numbers (indicating images) after they are no longer needed

Fix number of image tokens (#51)

Rewrite image embedding to remove the in-place op (#53)

Use Flash Attention for CLIP image encoder (#52)

Update README.md

clean_sample_code (#36)

Make flash attention configurable in user code (#26)

Add link to Phi-3 vision ONNX models

correct the paper reference (#25)

Update README.md (#14)

Update README

Update README

Update README

Update README

Update README

chore(root): Initial files upload.

initial commit