Upload README.md with huggingface_hub

README.md

@@ -32,13 +32,14 @@ for use with [WinSTT](https://github.com/dahshury/WinSTT).
 
 ## Precisions
 
-- `q4` — 4-bit (`MatMulNBitsQuantizer`, block_size=32, symmetric), ~2.2 GB. Recommended default.
+- `` (default) — fp32. Best accuracy; also the fastest on the CPU EP this model runs on.
+- `q4` — 4-bit (`MatMulNBitsQuantizer`, block_size=32, symmetric), ~2.2 GB.
 
-Only q4 is published. The model is CPU-only (its attention kernel is not DirectML-compatible) and the
-ONNX Runtime CPU EP up-casts fp16→fp32, so fp16 / q4f16 would be slower and larger with no benefit. The
-`export_merged.py` + `quantize.py` scripts here reproduce fp32 or any other precision on demand.
+fp16 / q4f16 are not shipped: the model is CPU-only (its attention kernel is not DirectML-compatible)
+and the ONNX Runtime CPU EP up-casts fp16→fp32, so they'd be slower and larger with no benefit.
 
 Verified numerically against the original PyTorch model (encoder/decoder parity ~1e-5 in fp32) and by
-end-to-end greedy decoding.
+end-to-end greedy decoding. See `export_merged.py` / `quantize.py` to reproduce.
+
 
 License: Apache-2.0 (inherited from the base model).