huggingface / candle

Minimalist ML framework for Rust
Apache License 2.0
13.79k stars 751 forks source link

Updated quantized phi model #2099

Closed LaurentMazare closed 3 weeks ago

EricLBuehler commented 3 weeks ago

Ah, that was what I was missing! I had issues in this PR when trying to break apart the qkv tensors, it seems so clear now. Thanks!