Next training iteration (1.28m)

added_tokens.json

@@ -1 +1 @@
-{"START_SELFPROMO_TOKEN": 32115, "PROFANITY_TOKEN": 32109, "NUMBER_TOKEN": 32103, "END_SELFPROMO_TOKEN": 32116, "NUMBER_PERCENTAGE_TOKEN": 32102, "START_INTERACTION_TOKEN": 32117, "[Laughter]": 32108, "END_INTERACTION_TOKEN": 32118, "END_SPONSOR_TOKEN": 32114, "[Music]": 32106, "URL_TOKEN": 32100, "SHORT_HYPHENATED_TOKEN": 32104, "[Applause]": 32107, "BETWEEN_SEGMENTS_TOKEN": 32111, "HYPHENATED_URL_TOKEN": 32101, "LONG_WORD_TOKEN": 32105, "START_SPONSOR_TOKEN": 32113, "EXTRACT_SEGMENTS: ": 32110, "NO_SEGMENT_TOKEN": 32112}
+{"NO_SEGMENT_TOKEN": 32112, "BETWEEN_SEGMENTS_TOKEN": 32111, "NUMBER_TOKEN": 32103, "START_SPONSOR_TOKEN": 32113, "START_SELFPROMO_TOKEN": 32115, "END_SPONSOR_TOKEN": 32114, "END_INTERACTION_TOKEN": 32118, "[Applause]": 32107, "HYPHENATED_URL_TOKEN": 32101, "EXTRACT_SEGMENTS: ": 32110, "SHORT_HYPHENATED_TOKEN": 32104, "END_SELFPROMO_TOKEN": 32116, "NUMBER_PERCENTAGE_TOKEN": 32102, "PROFANITY_TOKEN": 32109, "[Music]": 32106, "[Laughter]": 32108, "LONG_WORD_TOKEN": 32105, "URL_TOKEN": 32100, "START_INTERACTION_TOKEN": 32117}

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:106b3b3a93a21aec81b592c3f8a9f7a5312b94cdb7bb06045fe0295561836d60
+oid sha256:2d8a0af26f7ff16bae5ef0583063c69ee47700be6a4793c994a3ef6e83d117b8
 size 307892869

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:58daa822fb2c293f6f701f05ee612b3157bf523629556b4c5dba1950ca618f9b
+oid sha256:d8191c03018e0f166e6e407d71defd3cbfe5dcae86d404ff6a5903f7b765d8ac
 size 14503

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:64600c047defced1e13115233e81c861710facbdda4880101ca493398e58830f
+oid sha256:257198b9da6abece4de57c5b798cb130a628ec21c36fad1093e79d27e917992e
 size 623

trainer_state.json

@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 3.725608904205281,
-  "global_step": 1040000,
+  "epoch": 3.9405478794478936,
+  "global_step": 1100000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1046,11 +1046,71 @@
       "eval_samples_per_second": 87.844,
       "eval_steps_per_second": 21.962,
       "step": 1040000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.119283966627142e-05,
+      "loss": 0.0534,
+      "step": 1050000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1013723853569243e-05,
+      "loss": 0.0552,
+      "step": 1060000
+    },
+    {
+      "epoch": 3.8,
+      "eval_loss": 0.054923560470342636,
+      "eval_runtime": 716.4597,
+      "eval_samples_per_second": 86.583,
+      "eval_steps_per_second": 21.647,
+      "step": 1060000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.083460804086706e-05,
+      "loss": 0.055,
+      "step": 1070000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.065549222816489e-05,
+      "loss": 0.0547,
+      "step": 1080000
+    },
+    {
+      "epoch": 3.87,
+      "eval_loss": 0.05532824248075485,
+      "eval_runtime": 713.1552,
+      "eval_samples_per_second": 86.984,
+      "eval_steps_per_second": 21.747,
+      "step": 1080000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.047637641546271e-05,
+      "loss": 0.0526,
+      "step": 1090000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0297260602760537e-05,
+      "loss": 0.0536,
+      "step": 1100000
+    },
+    {
+      "epoch": 3.94,
+      "eval_loss": 0.05594659596681595,
+      "eval_runtime": 765.7749,
+      "eval_samples_per_second": 81.007,
+      "eval_steps_per_second": 20.253,
+      "step": 1100000
     }
   ],
   "max_steps": 2791490,
   "num_train_epochs": 10,
-  "total_flos": 6.949609736314368e+17,
+  "total_flos": 7.350678561491497e+17,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fdae48cf220f58bac313283bfdb731842739dd0b04d2098a35b1658df6a323f7
+oid sha256:01e951f41d1c838fb03d74c0997ade49e243ec0d07cb18ff2559ff234bae126c
 size 3119