Model save

README.md

@@ -60,12 +60,12 @@ The following hyperparameters were used during training:
 
 ### Training results
 
-| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
-|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
-| 0.5606        | 0.21  | 100  | 0.5444          | -0.6103        | -1.2336          | 0.7539             | 0.6232          | -380.4666      | -337.5370    | -2.9992         | -2.9936       |
-| 0.5399        | 0.42  | 200  | 0.5208          | -0.8033        | -1.6027          | 0.7773             | 0.7994          | -417.3810      | -356.8376    | -2.7196         | -2.7673       |
-| 0.5025        | 0.63  | 300  | 0.5035          | -0.9722        | -1.9486          | 0.7852             | 0.9764          | -451.9680      | -373.7211    | -2.1705         | -2.3401       |
-| 0.5005        | 0.84  | 400  | 0.4964          | -1.0116        | -2.0114          | 0.7969             | 0.9998          | -458.2532      | -377.6674    | -2.3544         | -2.4795       |
+| Training Loss | Epoch | Step | Logits/chosen | Logits/rejected | Logps/chosen | Logps/rejected | Validation Loss | Rewards/accuracies | Rewards/chosen | Rewards/margins | Rewards/rejected |
+|:-------------:|:-----:|:----:|:-------------:|:---------------:|:------------:|:--------------:|:---------------:|:------------------:|:--------------:|:---------------:|:----------------:|
+| 0.5606        | 0.21  | 100  | -2.9936       | -2.9992         | -337.5370    | -380.4666      | 0.5444          | 0.7539             | -0.6103        | 0.6232          | -1.2336          |
+| 0.5399        | 0.42  | 200  | -2.7673       | -2.7196         | -356.8376    | -417.3810      | 0.5208          | 0.7773             | -0.8033        | 0.7994          | -1.6027          |
+| 0.5025        | 0.63  | 300  | -2.3401       | -2.1705         | -373.7211    | -451.9680      | 0.5035          | 0.7852             | -0.9722        | 0.9764          | -1.9486          |
+| 0.5005        | 0.84  | 400  | -2.4795       | -2.3544         | -377.6674    | -458.2532      | 0.4964          | 0.7969             | -1.0116        | 0.9998          | -2.0114          |
 
 
 ### Framework versions

all_results.json

@@ -9,13 +9,13 @@
     "eval_rewards/chosen": -0.9225484132766724,
     "eval_rewards/margins": 1.0126549005508423,
     "eval_rewards/rejected": -1.9352033138275146,
-    "eval_runtime": 89.0705,
+    "eval_runtime": 86.9848,
     "eval_samples": 2000,
-    "eval_samples_per_second": 22.454,
-    "eval_steps_per_second": 0.359,
-    "train_loss": 0.5373976499964502,
-    "train_runtime": 5297.1261,
+    "eval_samples_per_second": 22.993,
+    "eval_steps_per_second": 0.368,
+    "train_loss": 0.08028295149863016,
+    "train_runtime": 773.4064,
     "train_samples": 61155,
-    "train_samples_per_second": 11.545,
-    "train_steps_per_second": 0.09
+    "train_samples_per_second": 79.072,
+    "train_steps_per_second": 0.618
 }

eval_results.json

@@ -9,8 +9,8 @@
     "eval_rewards/chosen": -0.9225484132766724,
     "eval_rewards/margins": 1.0126549005508423,
     "eval_rewards/rejected": -1.9352033138275146,
-    "eval_runtime": 89.0705,
+    "eval_runtime": 86.9848,
     "eval_samples": 2000,
-    "eval_samples_per_second": 22.454,
-    "eval_steps_per_second": 0.359
+    "eval_samples_per_second": 22.993,
+    "eval_steps_per_second": 0.368
 }

runs/Jan04_15-24-24_ip-26-0-172-73/events.out.tfevents.1704381944.ip-26-0-172-73.2335235.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:778321c06a661ef16a7dc9f3f9747bc041b58741a7cf7833b784efefb450e4ca
+size 9264

runs/Jan04_15-24-24_ip-26-0-172-73/events.out.tfevents.1704382805.ip-26-0-172-73.2335235.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6fe41322f218fef66fe99848864bfb75653d8fc2930784cf2fac3bdc7813ae2c
+size 828

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 1.0,
-    "train_loss": 0.5373976499964502,
-    "train_runtime": 5297.1261,
+    "train_loss": 0.08028295149863016,
+    "train_runtime": 773.4064,
     "train_samples": 61155,
-    "train_samples_per_second": 11.545,
-    "train_steps_per_second": 0.09
+    "train_samples_per_second": 79.072,
+    "train_steps_per_second": 0.618
 }

trainer_state.json

@@ -748,10 +748,10 @@
       "epoch": 1.0,
       "step": 478,
       "total_flos": 0.0,
-      "train_loss": 0.5373976499964502,
-      "train_runtime": 5297.1261,
-      "train_samples_per_second": 11.545,
-      "train_steps_per_second": 0.09
+      "train_loss": 0.08028295149863016,
+      "train_runtime": 773.4064,
+      "train_samples_per_second": 79.072,
+      "train_steps_per_second": 0.618
     }
   ],
   "logging_steps": 10,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f027d41f29be82338938a0b16542f35f8042f6098a904bca214b45fed5284b2c
+oid sha256:426b16da5d6c49955d8ee8f36eaf09fdcabfc25874f0ff18a47e145cecb00c63
 size 5944