Add Neural Metrics card (upstream card preserved below)

README.md

@@ -19,6 +19,30 @@ configs:
   - split: omnimath
     path: "omnimath.json"
 ---
+<div align="center">
+
+# Neural Metrics · Step-level evaluation, because extraction is a chain.
+
+<img src="https://img.shields.io/badge/Neural%20Metrics-document%20extraction-4F46E5?style=for-the-badge" alt="Neural Metrics" />
+<img src="https://img.shields.io/badge/fork%20of-Qwen%2FProcessBench-2563EB?style=flat-square" alt="fork" />
+
+</div>
+
+Document extraction is a multi-step process, and an end-to-end accuracy number hides *where* it broke. ProcessBench evaluates reasoning step by step, which is the shape of evaluation we want for our own pipelines.
+
+**We use it for:** benchmarking step-level error localization - studying how failures propagate through a chain rather than just counting final-answer mistakes.
+
+> ### Attribution
+> This is an **unmodified fork** of [`Qwen/ProcessBench`](https://huggingface.co/datasets/Qwen/ProcessBench), created by the [Qwen team](https://huggingface.co/Qwen).
+> All weights, files and behaviour are identical to upstream — we rehost it so our experiments stay
+> reproducible and version-pinned. The original license and all credit remain with the Qwen team.
+> If you want the canonical dataset, please use [the original](https://huggingface.co/datasets/Qwen/ProcessBench).
+
+---
+
+<details>
+<summary><b>Original dataset card from Qwen/ProcessBench</b> (click to expand)</summary>
+
 # ProcessBench
 
 This repository contains the dataset of the [ProcessBench](https://huggingface.co/papers/2412.06559) benchmark proposed by Qwen Team.
@@ -65,3 +89,5 @@ print(json.dumps(dataset[0], indent=2))
 }
 """
 ```
+
+</details>