Upload README.md with huggingface_hub

README.md

@@ -24,7 +24,6 @@ All variants take and return fp32 tensors — swap the `.pte` file, keep your ap
 |-----------|------|-----------|------------------------------------|------------------|
 | fp32 | `clip_vit_b32_image_xnnpack_fp32.pte` | 351.6 | 1.000000 | 19.0 |
 | fp16 | `clip_vit_b32_image_xnnpack_fp16.pte` | 180.7 | 0.999996 | 26.1 |
-| int8 (dynamic) | `clip_vit_b32_image_xnnpack_int8.pte` | 95.9 | 0.995739 | 18.4 |
 | Core ML (fp16, iOS) | `clip_vit_b32_image_coreml_all.pte` | 176.2 | 0.999998 | 3.5 |
 
 
@@ -40,9 +39,39 @@ only, not a device number (torch eager fp32 on the same machine: 18.5 ms).
 
 ### Checked in the task's own units
 
-Correlation is a first filter. These are the numbers that decide:
+Correlation is a first filter, and on an embedding it is a weak one: a build can read
+0.999 against eager and still move an answer, because what decides is whether the error is
+smaller than the gap between the answer a query gets and the runner-up.
 
-- **int8 (dynamic)** — measured in the units that matter for this model — cosine similarity of the image embeddings: median 0.9988 over 10 real images, worst 0.9957.
+The test is **near-duplicate retrieval** over 52 photographs, each also present centre-cropped to 90% and re-encoded as JPEG. Every item has one obviously
+correct nearest neighbour, and each build has to find it.
+
+| build | correct neighbour kept | worst score shift | budget spent |
+|---|---|---|---|
+| fp32 | 104 of 104 | 0.0000 | 0% |
+| fp16 | 104 of 104 | 0.0023 | 32% |
+| Core ML (fp16, iOS) | 104 of 104 | 0.0030 | 43% |
+
+The closest decision eager makes on this set is **0.0071**, and the bar is half of
+it. An earlier version of this card cleared int8 on "cosine similarity of the embeddings",
+which is correlation wearing a task metric's clothes — it never asked whether a retrieval
+answer moved.
+
+## Withdrawn: int8 (2026-08-27)
+
+`clip_vit_b32_image_xnnpack_int8.pte` was published and has been **withdrawn**. It was
+cleared on "cosine similarity of the image embeddings, median 0.9988 over 10 real images",
+which never asked whether a retrieval answer moved.
+
+Re-measured on the test above it still matches all 104 copies to their originals, but it
+moves a pair score by **0.0404** against a closest decision of **0.0071** — **567% of the
+room available**, against a bar of 50%. Its zero-shot label also moved on one image of the
+104, where no other build moved any. And the number is generous: this shelf calibrated the
+int8 build on these same photographs.
+
+At 95.9 MB it was the smallest build and size was the reason to reach for it. fp16 is
+180.7 MB and spends 32% of the same budget; the Core ML build is 176.2 MB, spends 43%, and
+runs 5x faster.
 
 ## Verification (executorch 1.4.0, torch 2.13.0)
 
@@ -63,13 +92,3 @@ torch.export -> to_edge_transform_and_lower(partitioner) -> .pte
 This repo holds **both towers**: `clip_vit_b32_image_xnnpack_fp32.pte` (image) and
 `clip_vit_b32_text_xnnpack_fp32.pte` (text, fixed len 77 + attention mask).
 L2-normalize both embeddings, then cosine-match.
-
-<!-- funnel:v1 -->
-
----
-
-**More models in this format:** [ExecuTorch Model Zoo](https://huggingface.co/collections/mlboydaisuke/executorch-model-zoo-6a7ff328390b63075ffeae5e) — 31 models, each with the recipe that produced it.
-
-**Want a different model on-device?** [Open a request](https://github.com/john-rocky/on-device-requests) — free, open weights only; the export and its measured numbers get published publicly.
-
-<!-- /funnel:v1 -->