Upload ray_tune_logs/params.json with huggingface_hub

ray_tune_logs/params.json

@@ -22,7 +22,7 @@
     "grad_norm": 0.5,
     "gradient_accumulation_steps": 1,
     "gradient_checkpoint": false,
-    "group_relative_norm": true,
+    "group_relative_norm": false,
     "inference_config": {
       "do_sample": true,
       "max_new_tokens": 250,
@@ -60,17 +60,17 @@
     "push_to_hub": null,
     "ref_role1_name_or_path": "gpt2",
     "ref_role2_name_or_path": "gpt2",
-    "rl_algo": "on_policy",
+    "rl_algo": "off_policy",
     "rl_w": 1.0,
     "rollout_config": {
-      "accuracy_w": 2.93586436439536,
-      "accuracy_w2": 59.56516018710569,
+      "accuracy_w": 0.03744666535741567,
+      "accuracy_w2": 60.07249475906205,
       "len_pen": 1.0,
       "len_pen2": 1.0,
-      "n_gen1": 2,
+      "n_gen1": 1,
       "n_gen2": 1,
       "similarity_fn": "rouge",
-      "threshold": 0.02310152225018238
+      "threshold": 0.011058510725696462
     },
     "rollout_game": "baseline3v2",
     "sample_config": {