Upload README.md with huggingface_hub

README.md

@@ -211,21 +211,25 @@ Measured on 400 benchmark rows against fp32, scoring **decision agreement** at t
 (the only metric that matters for a gate — a build can look fine on mean error and still flip
 calls near the boundary):
 
-| precision | decision agreement | max ΔP(deny) | memory |
-|---|---|---|---|
-| fp32 | reference | — | ~3.9 GB |
-| **fp16** | **100.00%** | 3.2e-03 | **~2 GB** |
-| bf16 | 99.75% | 2.7e-02 | ~2 GB |
-| int8 (ONNX dynamic) | 95.00% | 9.4e-01 | ~1 GB |
-
-**fp16 halves memory at zero measured cost** — pass `dtype=torch.float16` as shown above.
-
-Two results worth knowing. **fp16 beats bf16**: bf16 spends bits on exponent range an encoder
-forward pass does not need, trading away mantissa precision to get it — the opposite of the
-tradeoff you want at inference, even though bf16 is correct for training. And **int8 is not
-salvageable** — per-channel quantization, the standard fix, scored *worse* (94.25%). The failure
-is activation outliers in the GeGLU layers, which dynamic quantization cannot handle. See the
-[ONNX repo](https://huggingface.co/ProCreations/auto-1b-ONNX) for the full breakdown.
+Re-running the **full 3,000-item benchmark** at each precision:
+
+| precision | accuracy | AUROC | false-approve | memory |
+|---|---|---|---|---|
+| fp32 | 0.964000 | 0.992845 | 0.040685 | ~3.9 GB |
+| **bf16** | **0.964000** | **0.992845** | **0.040685** | **~2 GB** |
+| fp16 | 0.964333 | 0.992839 | 0.039971 | ~2 GB |
+| int8 (ONNX dynamic) | — | — | — | ~1 GB |
+
+**bf16 is exactly lossless** — identical on every metric and every context-length slice, zero
+flipped verdicts across 3,000 items. Ready-made at
+[`ProCreations/auto-1b-bf16`](https://huggingface.co/ProCreations/auto-1b-bf16), or pass
+`dtype=torch.bfloat16` here. fp16 differs by a single item with marginally worse AUROC; the two
+are equivalent in practice, and bf16 is preferred for carrying no overflow risk.
+
+**int8 is not salvageable.** It flips roughly 1 verdict in 20, and per-channel quantization —
+the standard fix — scored *worse* (94.25% vs 95.00% decision agreement). The failure is
+activation outliers in the GeGLU layers, which dynamic quantization cannot handle. See the
+[ONNX repo](https://huggingface.co/ProCreations/auto-1b-ONNX) for the breakdown.
 
 ## How it was built
 
@@ -262,5 +266,11 @@ benchmark is excluded by content hash (verified 0/3000 overlap).
 
 ## Other formats
 
-- [`ProCreations/auto-1b-gguf`](https://huggingface.co/ProCreations/auto-1b-gguf) — llama.cpp
-- [`ProCreations/auto-1b-ONNX`](https://huggingface.co/ProCreations/auto-1b-ONNX) — ONNX + int8
+- [`ProCreations/auto-1b-bf16`](https://huggingface.co/ProCreations/auto-1b-bf16) — half the
+  size, benchmark-identical to this model
+- [`ProCreations/auto-1b-ONNX`](https://huggingface.co/ProCreations/auto-1b-ONNX) — ONNX fp32
+  (its int8 build is documented as unfit for gating)
+
+A GGUF build was published and then **withdrawn**: llama.cpp converts the model, but its
+`--pooling rank` path returns zero for a 2-class classification head, so it could not actually
+make approve/deny decisions.