Upload exp_phase8_t3_rl_edit_tight_dapo_s1_20260506_083012/manifest.json with huggingface_hub

exp_phase8_t3_rl_edit_tight_dapo_s1_20260506_083012/manifest.json

@@ -0,0 +1,13 @@
+{
+  "trainer": "train_mdlm_rl_t3_motif.py",
+  "algo": "MDLM-GRPO with motif-triggered chained reward",
+  "alpha_oracle": 0.0,
+  "beta_motif": 1.0,
+  "gamma_jsd": 0.0,
+  "delta_kd": 0.0,
+  "gamma_step": 0.99,
+  "lora_r": 16,
+  "rollouts_per_prompt": 4,
+  "num_mdlm_steps": 8,
+  "num_rl_steps": 200
+}