Training in progress, step 100, checkpoint

checkpoint-100/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:66298e88cce2631c468d9284935e93b4d249f9dfa01080d4519bdf0479322d10
+oid sha256:9e47d738248a63d4e3f11ec8a3a89efe557eee3507f3bbb4eedddff0cdc82831
 size 54560368

checkpoint-100/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a75f97964133629dde1484c53985c085332965a689a7d706870c0f1e35d92853
+oid sha256:b8a53af0dcd9e12b7aec0a5757408a2c65451092aab2e12bbf6691d0edfb8206
 size 109267450

checkpoint-100/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d18870198c31c5821df8b9e9d648f47112d84cad0aef4ef7508fe2f35542f854
+oid sha256:12711e870dc2eb008cef779e781ac8fe1872c7e9d23f2bbc181148f8c2d8f981
 size 1064

checkpoint-100/trainer_state.json

@@ -1,6 +1,6 @@
 {
   "best_metric": 0.47333332896232605,
-  "best_model_checkpoint": "./zephyr/05-04-24-Weni-WeniGPT-Agents-Zephyr-1.0.1-KTO_testing kto dataset during training-3_max_steps-145_batch_16_2024-04-05_ppid_9/checkpoint-100",
+  "best_model_checkpoint": "./zephyr/06-04-24-Weni-WeniGPT-Agents-Zephyr-1.0.1-KTO_testing kto dataset during training-3_max_steps-145_batch_16_2024-04-06_ppid_9/checkpoint-100",
   "epoch": 0.684931506849315,
   "eval_steps": 50,
   "global_step": 100,
@@ -10,95 +10,95 @@
   "log_history": [
     {
       "epoch": 0.14,
-      "grad_norm": 32.5521125793457,
-      "kl": 6.568634033203125,
-      "learning_rate": 0.00018285714285714286,
-      "logps/chosen": -309.11578369140625,
-      "logps/rejected": -330.8851013183594,
-      "loss": 0.4463,
-      "rewards/chosen": -1.5149470567703247,
-      "rewards/margins": 0.730856716632843,
-      "rewards/rejected": -2.1397347450256348,
+      "grad_norm": 34.550357818603516,
+      "kl": 40.30364227294922,
+      "learning_rate": 0.00018571428571428572,
+      "logps/chosen": -341.70428466796875,
+      "logps/rejected": -376.15447998046875,
+      "loss": 0.4341,
+      "rewards/chosen": 4.142109394073486,
+      "rewards/margins": 5.559747695922852,
+      "rewards/rejected": -1.2980353832244873,
       "step": 20
     },
     {
       "epoch": 0.27,
-      "grad_norm": 5.194253444671631,
-      "kl": 9.558149337768555,
-      "learning_rate": 0.00015571428571428572,
-      "logps/chosen": -317.7120666503906,
-      "logps/rejected": -336.9992980957031,
-      "loss": 0.4204,
-      "rewards/chosen": -0.900913655757904,
-      "rewards/margins": 1.870524287223816,
-      "rewards/rejected": -2.7778050899505615,
+      "grad_norm": 0.0,
+      "kl": 9.54911994934082,
+      "learning_rate": 0.00015857142857142857,
+      "logps/chosen": -2304.16552734375,
+      "logps/rejected": -2427.63427734375,
+      "loss": 0.5152,
+      "rewards/chosen": -194.421875,
+      "rewards/margins": 7.0421857833862305,
+      "rewards/rejected": -205.4108428955078,
       "step": 40
     },
     {
       "epoch": 0.34,
       "eval_kl": 0.0,
-      "eval_logps/chosen": -471.27984619140625,
-      "eval_logps/rejected": -470.8447570800781,
-      "eval_loss": 0.47356343269348145,
-      "eval_rewards/chosen": -17.10463523864746,
-      "eval_rewards/margins": 1.9298614263534546,
-      "eval_rewards/rejected": -19.51471710205078,
-      "eval_runtime": 137.637,
-      "eval_samples_per_second": 2.18,
-      "eval_steps_per_second": 0.545,
+      "eval_logps/chosen": -3220.12841796875,
+      "eval_logps/rejected": -2905.742919921875,
+      "eval_loss": 0.47333332896232605,
+      "eval_rewards/chosen": -289.121337890625,
+      "eval_rewards/margins": -35.262752532958984,
+      "eval_rewards/rejected": -256.6285705566406,
+      "eval_runtime": 138.045,
+      "eval_samples_per_second": 2.173,
+      "eval_steps_per_second": 0.543,
       "step": 50
     },
     {
       "epoch": 0.41,
       "grad_norm": 0.0,
-      "kl": 0.2518434524536133,
-      "learning_rate": 0.00012714285714285714,
-      "logps/chosen": -638.1253051757812,
-      "logps/rejected": -641.34716796875,
-      "loss": 0.402,
-      "rewards/chosen": -33.93111038208008,
-      "rewards/margins": 1.04580819606781,
-      "rewards/rejected": -34.0406379699707,
+      "kl": 0.0,
+      "learning_rate": 0.00013000000000000002,
+      "logps/chosen": -3215.18505859375,
+      "logps/rejected": -3101.622802734375,
+      "loss": 0.4344,
+      "rewards/chosen": -290.7165222167969,
+      "rewards/margins": -12.484609603881836,
+      "rewards/rejected": -276.9840393066406,
       "step": 60
     },
     {
       "epoch": 0.55,
       "grad_norm": 0.0,
       "kl": 0.0,
-      "learning_rate": 9.857142857142858e-05,
-      "logps/chosen": -1467.325927734375,
-      "logps/rejected": -1537.5704345703125,
-      "loss": 0.45,
-      "rewards/chosen": -117.26831817626953,
-      "rewards/margins": 9.93100357055664,
-      "rewards/rejected": -124.3635025024414,
+      "learning_rate": 0.00010142857142857143,
+      "logps/chosen": -3785.22607421875,
+      "logps/rejected": -3411.793701171875,
+      "loss": 0.4781,
+      "rewards/chosen": -344.2449951171875,
+      "rewards/margins": -43.561405181884766,
+      "rewards/rejected": -306.4319152832031,
       "step": 80
     },
     {
       "epoch": 0.68,
       "grad_norm": 0.0,
       "kl": 0.0,
-      "learning_rate": 7e-05,
-      "logps/chosen": -2025.484375,
-      "logps/rejected": -1882.913818359375,
-      "loss": 0.4813,
-      "rewards/chosen": -169.48184204101562,
-      "rewards/margins": -10.249415397644043,
-      "rewards/rejected": -157.5347900390625,
+      "learning_rate": 7.285714285714286e-05,
+      "logps/chosen": -3814.2275390625,
+      "logps/rejected": -3579.71630859375,
+      "loss": 0.4344,
+      "rewards/chosen": -343.0238952636719,
+      "rewards/margins": -25.151195526123047,
+      "rewards/rejected": -323.59967041015625,
       "step": 100
     },
     {
       "epoch": 0.68,
       "eval_kl": 0.0,
-      "eval_logps/chosen": -1703.4486083984375,
-      "eval_logps/rejected": -1636.100341796875,
+      "eval_logps/chosen": -3431.77734375,
+      "eval_logps/rejected": -3077.673583984375,
       "eval_loss": 0.47333332896232605,
-      "eval_rewards/chosen": -140.32151794433594,
-      "eval_rewards/margins": -6.890669822692871,
-      "eval_rewards/rejected": -136.040283203125,
-      "eval_runtime": 137.631,
-      "eval_samples_per_second": 2.18,
-      "eval_steps_per_second": 0.545,
+      "eval_rewards/chosen": -310.2862854003906,
+      "eval_rewards/margins": -38.59312438964844,
+      "eval_rewards/rejected": -273.8216552734375,
+      "eval_runtime": 138.0016,
+      "eval_samples_per_second": 2.174,
+      "eval_steps_per_second": 0.543,
       "step": 100
     }
   ],

checkpoint-100/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2537b1649ac52556ff9ceed8db71e4fc060d1ffc901556fa0b90e1d0d70234d5
+oid sha256:c4bc7e345e9ee96ec54aebcd4c5b76c26861e98332bb42d8866efd5772351c4a
 size 5624