Upload PolyGuard artifact: promoted qwen-qwen2-5-3b-instruct

outputs/reports/grpo_trl_run.json

@@ -0,0 +1,43 @@
+{
+  "status": "ok",
+  "backend": "trl_transformers",
+  "model_id": "Qwen/Qwen2.5-3B-Instruct",
+  "records": 2000,
+  "prompts_path": "/app/data/processed/training_corpus_grpo_prompts.jsonl",
+  "reward_summary": {
+    "count": 4000,
+    "avg_reward": 0.767,
+    "avg_reward_components": {
+      "format_compliance_score": 0.999,
+      "candidate_alignment_score": 0.999,
+      "legality_score": 0.929,
+      "safety_delta_score": 0.497,
+      "burden_improvement_score": 0.469,
+      "disease_stability_score": 0.861,
+      "dosing_quality_score": 0.526,
+      "abstention_quality_score": 0.56,
+      "efficiency_score": 0.849,
+      "process_fidelity_score": 0.856,
+      "explanation_grounding_score": 0.795,
+      "anti_cheat_score": 0.589,
+      "uncertainty_calibration_score": 0.747
+    },
+    "avg_primary_reward_channels": {
+      "safety_legality": 0.816,
+      "clinical_improvement": 0.609,
+      "dosing_quality": 0.543,
+      "process_integrity": 0.875
+    }
+  },
+  "reward_log": "/app/checkpoints/sweeps/qwen-qwen2-5-3b-instruct/grpo_reward_components.jsonl",
+  "train_metrics": {
+    "train_runtime": 6873.9375,
+    "train_samples_per_second": 0.291,
+    "train_steps_per_second": 0.291,
+    "total_flos": 0.0,
+    "train_loss": 2.665005830824185e-06
+  },
+  "history_path": "/app/checkpoints/sweeps/qwen-qwen2-5-3b-instruct/grpo_history.json",
+  "artifact_path": "/app/checkpoints/sweeps/qwen-qwen2-5-3b-instruct/grpo_adapter",
+  "unsloth_available": false
+}