Phi-4-mini-instruct — LiteRT-LM INT4
Local conversion of microsoft/Phi-4-mini-instruct to .litertlm (blockwise INT4) for on-device use with LiteRT-LM / expo-ai-kit.
| File | Quantization | Size | Context |
|---|---|---|---|
phi4-mini-int4.litertlm |
int4 (BOCTAV4 / block32) | ~2.6 GB | 4096 |
start_tokenremoved (Phi instruct has no BOS; required for correct chat)- Embeddings externalized for iOS section limits
- License: MIT (inherited from base model)
sha256(phi4-mini-int4.litertlm) = e7729c2213877109c56af345a5d1deb7a366c581e1515414985b8886fb6ac201
- Downloads last month
- 26
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for kloodia/Phi-4-mini-instruct-LiteRT-INT4
Base model
microsoft/Phi-4-mini-instruct