Add model card and metadata

README.md

@@ -1 +1,27 @@
-arxiv.org/abs/2603.27027
+---
+license: apache-2.0
+library_name: transformers
+pipeline_tag: text-generation
+---
+
+# TAPS: Task-Aware Proposal Distributions for Speculative Sampling
+
+This repository contains the weights for a draft model introduced in the paper [TAPS: Task Aware Proposal Distributions for Speculative Sampling](https://arxiv.org/abs/2603.27027).
+
+TAPS (Task-Aware Proposal Distributions for Speculative Sampling) investigates how the training distribution of draft models affects the performance of speculative decoding. The study demonstrates that specialized drafters (like HASS and EAGLE-2) trained on specific domains (such as math or chat) significantly improve acceptance length when matched with the downstream workload.
+
+- **Paper:** [TAPS: Task Aware Proposal Distributions for Speculative Sampling](https://arxiv.org/abs/2603.27027)
+- **GitHub Repository:** [Moe-Zbeeb/TAPS](https://github.com/Moe-Zbeeb/TAPS)
+
+## Model Description
+This specific checkpoint is a lightweight LLaMA-style drafter (~0.8B parameters) featuring a single hidden layer. It is designed to be used in conjunction with a larger target model (like Meta-Llama-3-8B-Instruct) to accelerate autoregressive generation via speculative decoding.
+
+## Citation
+```bibtex
+@article{zbib2026taps,
+  title={TAPS: Task Aware Proposal Distributions for Speculative Sampling},
+  author={Zbib, Mohamad and Bazzi, Mohamad and Mohanna, Ammar and Ghanem, Bernard and Hammoud, Hasan Abed Al Kader},
+  year={2026},
+  note={Technical report}
+}
+```