Add DexPolicy model card

README.md

@@ -0,0 +1,23 @@
+---
+tags:
+- reinforcement-learning
+- robotics
+- dexterous-manipulation
+---
+
+# DexPolicy representative checkpoints
+
+This repository contains a representative subset of the object-specific policies reported in the DexPolicy paper. The released pair uses the YCB mustard-bottle relocation task and isolates the paper's exploration-scheduling intervention under GRPO.
+
+## Checkpoints
+
+| File | Method | Object | Seed | Continuation |
+|---|---|---|---:|---|
+| `models/grpo_mustard_baseline_seed0.zip` | GRPO baseline | YCB 006 mustard bottle | 0 | Inherited, learned policy standard deviation |
+| `models/grpo_mustard_dexpolicy_seed0.zip` | GRPO + DexPolicy | YCB 006 mustard bottle | 0 | Scheduled standard deviation, `0.10 -> 0.03` |
+
+Both policies continue from matched object-and-seed checkpoints at 507,904 environment steps for another 250,000 steps. They use the same reward, terminal success definition, curriculum stage, and GRPO return-to-go credit module; the exploration schedule is the controlled difference.
+
+The archives follow the Stable-Baselines3 checkpoint layout and include the policy, PyTorch variables, optimizer state, and library-version metadata. These policies are object-specific and are not presented as a single cross-object policy.
+
+This is a compact representative release rather than the complete multi-object checkpoint suite.