Training in progress, step 1500

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a83d23f9c280b1b78545a720da15a8e9067a3694eaf73c8bd335a73ff1dfdd14
+oid sha256:b03bffd09b24ec13e3096504399449cf97bf57621524dc0504b7d091fab80fdd
 size 268290900

run-2/checkpoint-1500/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b9cf0f8a2189351356d6987a53766ff1ba02473ec638516dde50f31a1e8179fd
+oid sha256:b03bffd09b24ec13e3096504399449cf97bf57621524dc0504b7d091fab80fdd
 size 268290900

run-2/checkpoint-1500/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7582495c2bb697085fde7d23892867d0d72d934021d5a2f5f7a9fe88f3833167
+oid sha256:9e5c6bf52e7a2bfb516bde6af6e76d5ca6dbc8e7f9511bf5096c4a7a88abd81f
 size 536643898

run-2/checkpoint-1500/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ad4d7d251acf36e559c362893a1fb310c9f46b20e8a330025a14b6829ce4ab07
+oid sha256:55c8d3ce0734337fc0c187ca5543b4c70ca45d996531f199209b3a0c2a798109
 size 1064

run-2/checkpoint-1500/trainer_state.json

@@ -10,68 +10,68 @@
   "log_history": [
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.5851612903225807,
-      "eval_loss": 0.2035684436559677,
-      "eval_runtime": 5.6989,
-      "eval_samples_per_second": 543.962,
-      "eval_steps_per_second": 11.406,
+      "eval_accuracy": 0.6,
+      "eval_loss": 0.22239726781845093,
+      "eval_runtime": 5.4481,
+      "eval_samples_per_second": 569.006,
+      "eval_steps_per_second": 11.931,
       "step": 318
     },
     {
       "epoch": 1.57,
-      "learning_rate": 1.550763701707098e-05,
-      "loss": 0.3221,
+      "learning_rate": 1.4758909853249476e-05,
+      "loss": 0.3484,
       "step": 500
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.8335483870967741,
-      "eval_loss": 0.10106482356786728,
-      "eval_runtime": 5.8785,
-      "eval_samples_per_second": 527.345,
-      "eval_steps_per_second": 11.057,
+      "eval_accuracy": 0.83,
+      "eval_loss": 0.109930619597435,
+      "eval_runtime": 5.7104,
+      "eval_samples_per_second": 542.874,
+      "eval_steps_per_second": 11.383,
       "step": 636
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.885483870967742,
-      "eval_loss": 0.06870077550411224,
-      "eval_runtime": 5.783,
-      "eval_samples_per_second": 536.053,
-      "eval_steps_per_second": 11.24,
+      "eval_accuracy": 0.8809677419354839,
+      "eval_loss": 0.07449585199356079,
+      "eval_runtime": 5.3903,
+      "eval_samples_per_second": 575.109,
+      "eval_steps_per_second": 12.059,
       "step": 954
     },
     {
       "epoch": 3.14,
-      "learning_rate": 1.101527403414196e-05,
-      "loss": 0.1162,
+      "learning_rate": 9.517819706498952e-06,
+      "loss": 0.1265,
       "step": 1000
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.8980645161290323,
-      "eval_loss": 0.053607575595378876,
-      "eval_runtime": 5.8706,
-      "eval_samples_per_second": 528.052,
-      "eval_steps_per_second": 11.072,
+      "eval_accuracy": 0.8964516129032258,
+      "eval_loss": 0.05887645110487938,
+      "eval_runtime": 5.4647,
+      "eval_samples_per_second": 567.278,
+      "eval_steps_per_second": 11.895,
       "step": 1272
     },
     {
       "epoch": 4.72,
-      "learning_rate": 6.522911051212939e-06,
-      "loss": 0.0765,
+      "learning_rate": 4.276729559748428e-06,
+      "loss": 0.0843,
       "step": 1500
     }
   ],
   "logging_steps": 500,
-  "max_steps": 2226,
-  "num_train_epochs": 7,
+  "max_steps": 1908,
+  "num_train_epochs": 6,
   "save_steps": 500,
   "total_flos": 389479376069112.0,
   "trial_name": null,
   "trial_params": {
-    "alpha": 0.8305833099612083,
-    "num_train_epochs": 7,
-    "temperature": 12
+    "alpha": 0.339174080203406,
+    "num_train_epochs": 6,
+    "temperature": 7
   }
 }

run-2/checkpoint-1500/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9c739c287048413530d77539873d77810678560a3c9c21cac037ba107d1ab725
+oid sha256:7bcd9576316ea4b78c0edb02ef77b7b54c380e6454ae67bc0bbd59e18e13a200
 size 4664

runs/Jan25_17-40-54_c146da53f02f/events.out.tfevents.1706206065.c146da53f02f.3236.3

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f24128428e8f274ac37e8f5de24cca3e5f330964e2cd9cca05f88629521eaf78
-size 13447
+oid sha256:6eb7aa29e29f29790ce6a5253807cf5aee0d4a8b0472e9e547949dbf83a42f7c
+size 14604