docs: mention AGENTS.md for LLM tools, add benchmark Space section
Browse files
README.md
CHANGED
|
@@ -25,7 +25,7 @@ Vision Transformer (ViT-Small) trained on 2.7M samples across 4,803 generators f
|
|
| 25 |
|
| 26 |
**If you downloaded this model before July 22, 2026, your local copy has incorrect config and weights.** Apologies for the mess — this model was originally hastily put together as an internal proof-of-concept for a hackathon, and we never imagined it would quietly become one of the top image classification models on Hugging Face. This update is long overdue.
|
| 27 |
|
| 28 |
-
The `model.safetensors` has been regenerated from the correct training checkpoint and all metadata has been fixed. For a detailed breakdown of every change, see [CHANGELOG.md](CHANGELOG.md).
|
| 29 |
|
| 30 |
| Bug | Effect | Fixed Value |
|
| 31 |
|---|---|---|
|
|
@@ -142,6 +142,27 @@ fake_prob = 1 / (1 + np.exp(-logit))
|
|
| 142 |
|
| 143 |
---
|
| 144 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 145 |
## Model Details
|
| 146 |
|
| 147 |
- **Developed by**: Jeongsoo Park and Andrew Owens, University of Michigan
|
|
|
|
| 25 |
|
| 26 |
**If you downloaded this model before July 22, 2026, your local copy has incorrect config and weights.** Apologies for the mess — this model was originally hastily put together as an internal proof-of-concept for a hackathon, and we never imagined it would quietly become one of the top image classification models on Hugging Face. This update is long overdue.
|
| 27 |
|
| 28 |
+
The `model.safetensors` has been regenerated from the correct training checkpoint and all metadata has been fixed. For a detailed breakdown of every change, see [CHANGELOG.md](CHANGELOG.md). If you use LLM-based coding agents (Claude Code, Cursor, GitHub Copilot, etc.), the repo includes an [AGENTS.md](AGENTS.md) to help your agent ramp up quickly.
|
| 29 |
|
| 30 |
| Bug | Effect | Fixed Value |
|
| 31 |
|---|---|---|
|
|
|
|
| 142 |
|
| 143 |
---
|
| 144 |
|
| 145 |
+
## Benchmark & Comparison Space
|
| 146 |
+
|
| 147 |
+
A companion Gradio Space lets you test every variant side by side — upload your own images and compare PyTorch vs ONNX performance in real time.
|
| 148 |
+
|
| 149 |
+
**What it does:**
|
| 150 |
+
|
| 151 |
+
| Tab | Description |
|
| 152 |
+
|---|---|
|
| 153 |
+
| **Compare** | Upload a single image, see PyTorch and all selected ONNX variants side by side with timing |
|
| 154 |
+
| **Benchmark** | Upload multiple images for batch processing, compare inference speed across all variants |
|
| 155 |
+
| **Help** | Variant selection guide and preprocessing details |
|
| 156 |
+
|
| 157 |
+
**Use it to:**
|
| 158 |
+
- See how quantization affects prediction confidence on your own images
|
| 159 |
+
- Measure real-world inference speed across variants (CPU/GPU)
|
| 160 |
+
- Verify the corrected model produces results consistent with the original timm pipeline
|
| 161 |
+
|
| 162 |
+
> Link coming soon — deploying as a separate Space. Follow the repo for updates.
|
| 163 |
+
|
| 164 |
+
---
|
| 165 |
+
|
| 166 |
## Model Details
|
| 167 |
|
| 168 |
- **Developed by**: Jeongsoo Park and Andrew Owens, University of Michigan
|