Training in progress, epoch 2, checkpoint

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:91e0796e5d9d9bc45508498937ca534640f6311a53399d4644caaa7cf9e44170
+oid sha256:c5ed1e7a0e3c59f320b1715a05c3666ecc2fc6c228708e3d959cb6460bc491a3
 size 74542452

last-checkpoint/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7c117429faf456604fc45db63d7a3e6d2c0e9780e29554d2b47051b9593f8a05
+oid sha256:b36bcb0fed2e275887daa41022406c825f99b4b86a41b59934d97ac489910923
 size 15984

last-checkpoint/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f105fb08fa0ecd11f3978f691ef2538e9e38abcac826bc9a4cfc09b26e29b39d
+oid sha256:274e4dab9a36d014099ff4e09521694810050a7d0e4b9f78328955bba8ef6b26
 size 15984

last-checkpoint/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:46cadbd0037afd7891b2504363982b19551511cc8ab68b7eb935c546bdf5f257
+oid sha256:29259f0bc54246fc694a705960e66e905a1b8b55b02460ff05ae4445d414a2d9
 size 15984

last-checkpoint/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f29b1295a826abfe9eb85474df78753ec1a813fb629c4d3ba10cf35df6e85a4c
+oid sha256:814a16321827f315a53372838b09d10fefd4b8ffca48166ac25206f574e28035
 size 15984

last-checkpoint/rng_state_4.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f53e106aae6177f40d43aed856e7b8b250fc9083bbb03799c6ae5e1b1aa38bdf
+oid sha256:81f6a1d71656a2ff79e6f54eb26413cd8443bae9f9540780db5c4f003dfc02c8
 size 15984

last-checkpoint/rng_state_5.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:20aa54d3cb811c77bc6fff62a7f06bfa974cb576d2702d38905fce17ae64c96e
+oid sha256:c77edaf995b7c7a38c1c51228d68765c7f1753a648cd2d0a15bbb24e4eb6ce2c
 size 15984

last-checkpoint/rng_state_6.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c0e0c579d0c670928190c3849f21b85955c12662303a7f4db75b8e6c4de069fb
+oid sha256:09d53a034ddeb077449f1e8333121b956bc22be6592d1e77823724a2c80d0744
 size 15984

last-checkpoint/rng_state_7.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:99feaa4bcbd2c03e964fb405aa2bb8c8116675ad27cfe0189cef2ec0b0bf8443
+oid sha256:ada87abc687bb97192f1629a03d9a5c57963c2beafcd71038e81f82064321897
 size 15984

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f37204616e16924b89b5010370d089acd6b6c4f8af3db2c929bd809019b481ea
+oid sha256:12e14872b017d468ccfc628bb0cb686bc840535b42756036bb777de9399606c3
 size 1064

last-checkpoint/trainer_state.json

@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.0,
+  "epoch": 2.9972602739726026,
   "eval_steps": 500,
-  "global_step": 365,
+  "global_step": 547,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -2578,6 +2578,1288 @@
       "eval_samples_per_second": 69.495,
       "eval_steps_per_second": 2.205,
       "step": 365
+    },
+    {
+      "epoch": 2.0054794520547947,
+      "grad_norm": NaN,
+      "learning_rate": 8.99638285148785e-05,
+      "loss": 0.0,
+      "step": 366
+    },
+    {
+      "epoch": 2.010958904109589,
+      "grad_norm": NaN,
+      "learning_rate": 8.933888457247402e-05,
+      "loss": 0.0,
+      "step": 367
+    },
+    {
+      "epoch": 2.0164383561643837,
+      "grad_norm": NaN,
+      "learning_rate": 8.871436151265184e-05,
+      "loss": 0.0,
+      "step": 368
+    },
+    {
+      "epoch": 2.021917808219178,
+      "grad_norm": NaN,
+      "learning_rate": 8.809028399051302e-05,
+      "loss": 0.0,
+      "step": 369
+    },
+    {
+      "epoch": 2.0273972602739727,
+      "grad_norm": NaN,
+      "learning_rate": 8.746667664356956e-05,
+      "loss": 0.0,
+      "step": 370
+    },
+    {
+      "epoch": 2.032876712328767,
+      "grad_norm": NaN,
+      "learning_rate": 8.684356409077176e-05,
+      "loss": 0.0,
+      "step": 371
+    },
+    {
+      "epoch": 2.0383561643835617,
+      "grad_norm": NaN,
+      "learning_rate": 8.62209709315362e-05,
+      "loss": 0.0,
+      "step": 372
+    },
+    {
+      "epoch": 2.043835616438356,
+      "grad_norm": NaN,
+      "learning_rate": 8.559892174477479e-05,
+      "loss": 0.0,
+      "step": 373
+    },
+    {
+      "epoch": 2.0493150684931507,
+      "grad_norm": NaN,
+      "learning_rate": 8.497744108792429e-05,
+      "loss": 0.0,
+      "step": 374
+    },
+    {
+      "epoch": 2.0547945205479454,
+      "grad_norm": NaN,
+      "learning_rate": 8.435655349597689e-05,
+      "loss": 0.0,
+      "step": 375
+    },
+    {
+      "epoch": 2.0602739726027397,
+      "grad_norm": NaN,
+      "learning_rate": 8.373628348051165e-05,
+      "loss": 0.0,
+      "step": 376
+    },
+    {
+      "epoch": 2.0657534246575344,
+      "grad_norm": NaN,
+      "learning_rate": 8.311665552872662e-05,
+      "loss": 0.0,
+      "step": 377
+    },
+    {
+      "epoch": 2.0712328767123287,
+      "grad_norm": NaN,
+      "learning_rate": 8.249769410247239e-05,
+      "loss": 0.0,
+      "step": 378
+    },
+    {
+      "epoch": 2.0767123287671234,
+      "grad_norm": NaN,
+      "learning_rate": 8.187942363728625e-05,
+      "loss": 0.0,
+      "step": 379
+    },
+    {
+      "epoch": 2.0821917808219177,
+      "grad_norm": NaN,
+      "learning_rate": 8.126186854142752e-05,
+      "loss": 0.0,
+      "step": 380
+    },
+    {
+      "epoch": 2.0876712328767124,
+      "grad_norm": NaN,
+      "learning_rate": 8.064505319491398e-05,
+      "loss": 0.0,
+      "step": 381
+    },
+    {
+      "epoch": 2.0931506849315067,
+      "grad_norm": NaN,
+      "learning_rate": 8.002900194855932e-05,
+      "loss": 0.0,
+      "step": 382
+    },
+    {
+      "epoch": 2.0986301369863014,
+      "grad_norm": NaN,
+      "learning_rate": 7.941373912301189e-05,
+      "loss": 0.0,
+      "step": 383
+    },
+    {
+      "epoch": 2.1041095890410957,
+      "grad_norm": NaN,
+      "learning_rate": 7.879928900779456e-05,
+      "loss": 0.0,
+      "step": 384
+    },
+    {
+      "epoch": 2.1095890410958904,
+      "grad_norm": NaN,
+      "learning_rate": 7.818567586034577e-05,
+      "loss": 0.0,
+      "step": 385
+    },
+    {
+      "epoch": 2.115068493150685,
+      "grad_norm": NaN,
+      "learning_rate": 7.75729239050619e-05,
+      "loss": 0.0,
+      "step": 386
+    },
+    {
+      "epoch": 2.1205479452054794,
+      "grad_norm": NaN,
+      "learning_rate": 7.696105733234098e-05,
+      "loss": 0.0,
+      "step": 387
+    },
+    {
+      "epoch": 2.126027397260274,
+      "grad_norm": NaN,
+      "learning_rate": 7.635010029762756e-05,
+      "loss": 0.0,
+      "step": 388
+    },
+    {
+      "epoch": 2.1315068493150684,
+      "grad_norm": NaN,
+      "learning_rate": 7.574007692045928e-05,
+      "loss": 0.0,
+      "step": 389
+    },
+    {
+      "epoch": 2.136986301369863,
+      "grad_norm": NaN,
+      "learning_rate": 7.513101128351454e-05,
+      "loss": 0.0,
+      "step": 390
+    },
+    {
+      "epoch": 2.1424657534246574,
+      "grad_norm": NaN,
+      "learning_rate": 7.45229274316618e-05,
+      "loss": 0.0,
+      "step": 391
+    },
+    {
+      "epoch": 2.147945205479452,
+      "grad_norm": NaN,
+      "learning_rate": 7.391584937101033e-05,
+      "loss": 0.0,
+      "step": 392
+    },
+    {
+      "epoch": 2.1534246575342464,
+      "grad_norm": NaN,
+      "learning_rate": 7.330980106796246e-05,
+      "loss": 0.0,
+      "step": 393
+    },
+    {
+      "epoch": 2.158904109589041,
+      "grad_norm": NaN,
+      "learning_rate": 7.270480644826749e-05,
+      "loss": 0.0,
+      "step": 394
+    },
+    {
+      "epoch": 2.1643835616438354,
+      "grad_norm": NaN,
+      "learning_rate": 7.210088939607708e-05,
+      "loss": 0.0,
+      "step": 395
+    },
+    {
+      "epoch": 2.16986301369863,
+      "grad_norm": NaN,
+      "learning_rate": 7.149807375300239e-05,
+      "loss": 0.0,
+      "step": 396
+    },
+    {
+      "epoch": 2.175342465753425,
+      "grad_norm": NaN,
+      "learning_rate": 7.089638331717284e-05,
+      "loss": 0.0,
+      "step": 397
+    },
+    {
+      "epoch": 2.180821917808219,
+      "grad_norm": NaN,
+      "learning_rate": 7.029584184229653e-05,
+      "loss": 0.0,
+      "step": 398
+    },
+    {
+      "epoch": 2.186301369863014,
+      "grad_norm": NaN,
+      "learning_rate": 6.969647303672262e-05,
+      "loss": 0.0,
+      "step": 399
+    },
+    {
+      "epoch": 2.191780821917808,
+      "grad_norm": NaN,
+      "learning_rate": 6.909830056250527e-05,
+      "loss": 0.0,
+      "step": 400
+    },
+    {
+      "epoch": 2.197260273972603,
+      "grad_norm": NaN,
+      "learning_rate": 6.850134803446954e-05,
+      "loss": 0.0,
+      "step": 401
+    },
+    {
+      "epoch": 2.202739726027397,
+      "grad_norm": NaN,
+      "learning_rate": 6.790563901927907e-05,
+      "loss": 0.0,
+      "step": 402
+    },
+    {
+      "epoch": 2.208219178082192,
+      "grad_norm": NaN,
+      "learning_rate": 6.731119703450577e-05,
+      "loss": 0.0,
+      "step": 403
+    },
+    {
+      "epoch": 2.213698630136986,
+      "grad_norm": NaN,
+      "learning_rate": 6.671804554770135e-05,
+      "loss": 0.0,
+      "step": 404
+    },
+    {
+      "epoch": 2.219178082191781,
+      "grad_norm": NaN,
+      "learning_rate": 6.612620797547087e-05,
+      "loss": 0.0,
+      "step": 405
+    },
+    {
+      "epoch": 2.2246575342465755,
+      "grad_norm": NaN,
+      "learning_rate": 6.55357076825483e-05,
+      "loss": 0.0,
+      "step": 406
+    },
+    {
+      "epoch": 2.23013698630137,
+      "grad_norm": NaN,
+      "learning_rate": 6.494656798087412e-05,
+      "loss": 0.0,
+      "step": 407
+    },
+    {
+      "epoch": 2.2356164383561645,
+      "grad_norm": NaN,
+      "learning_rate": 6.435881212867493e-05,
+      "loss": 0.0,
+      "step": 408
+    },
+    {
+      "epoch": 2.241095890410959,
+      "grad_norm": NaN,
+      "learning_rate": 6.377246332954544e-05,
+      "loss": 0.0,
+      "step": 409
+    },
+    {
+      "epoch": 2.2465753424657535,
+      "grad_norm": NaN,
+      "learning_rate": 6.318754473153221e-05,
+      "loss": 0.0,
+      "step": 410
+    },
+    {
+      "epoch": 2.252054794520548,
+      "grad_norm": NaN,
+      "learning_rate": 6.260407942621998e-05,
+      "loss": 0.0,
+      "step": 411
+    },
+    {
+      "epoch": 2.2575342465753425,
+      "grad_norm": NaN,
+      "learning_rate": 6.20220904478199e-05,
+      "loss": 0.0,
+      "step": 412
+    },
+    {
+      "epoch": 2.263013698630137,
+      "grad_norm": NaN,
+      "learning_rate": 6.144160077226036e-05,
+      "loss": 0.0,
+      "step": 413
+    },
+    {
+      "epoch": 2.2684931506849315,
+      "grad_norm": NaN,
+      "learning_rate": 6.086263331627976e-05,
+      "loss": 0.0,
+      "step": 414
+    },
+    {
+      "epoch": 2.2739726027397262,
+      "grad_norm": NaN,
+      "learning_rate": 6.0285210936521955e-05,
+      "loss": 0.0,
+      "step": 415
+    },
+    {
+      "epoch": 2.2794520547945205,
+      "grad_norm": NaN,
+      "learning_rate": 5.9709356428633746e-05,
+      "loss": 0.0,
+      "step": 416
+    },
+    {
+      "epoch": 2.2849315068493152,
+      "grad_norm": NaN,
+      "learning_rate": 5.913509252636511e-05,
+      "loss": 0.0,
+      "step": 417
+    },
+    {
+      "epoch": 2.2904109589041095,
+      "grad_norm": NaN,
+      "learning_rate": 5.856244190067159e-05,
+      "loss": 0.0,
+      "step": 418
+    },
+    {
+      "epoch": 2.2958904109589042,
+      "grad_norm": NaN,
+      "learning_rate": 5.799142715881938e-05,
+      "loss": 0.0,
+      "step": 419
+    },
+    {
+      "epoch": 2.3013698630136985,
+      "grad_norm": NaN,
+      "learning_rate": 5.7422070843492734e-05,
+      "loss": 0.0,
+      "step": 420
+    },
+    {
+      "epoch": 2.3068493150684932,
+      "grad_norm": NaN,
+      "learning_rate": 5.6854395431904094e-05,
+      "loss": 0.0,
+      "step": 421
+    },
+    {
+      "epoch": 2.3123287671232875,
+      "grad_norm": NaN,
+      "learning_rate": 5.6288423334906735e-05,
+      "loss": 0.0,
+      "step": 422
+    },
+    {
+      "epoch": 2.317808219178082,
+      "grad_norm": NaN,
+      "learning_rate": 5.572417689610987e-05,
+      "loss": 0.0,
+      "step": 423
+    },
+    {
+      "epoch": 2.323287671232877,
+      "grad_norm": NaN,
+      "learning_rate": 5.5161678390996796e-05,
+      "loss": 0.0,
+      "step": 424
+    },
+    {
+      "epoch": 2.328767123287671,
+      "grad_norm": NaN,
+      "learning_rate": 5.4600950026045326e-05,
+      "loss": 0.0,
+      "step": 425
+    },
+    {
+      "epoch": 2.334246575342466,
+      "grad_norm": NaN,
+      "learning_rate": 5.404201393785122e-05,
+      "loss": 0.0,
+      "step": 426
+    },
+    {
+      "epoch": 2.33972602739726,
+      "grad_norm": NaN,
+      "learning_rate": 5.348489219225416e-05,
+      "loss": 0.0,
+      "step": 427
+    },
+    {
+      "epoch": 2.345205479452055,
+      "grad_norm": NaN,
+      "learning_rate": 5.292960678346675e-05,
+      "loss": 0.0,
+      "step": 428
+    },
+    {
+      "epoch": 2.350684931506849,
+      "grad_norm": NaN,
+      "learning_rate": 5.237617963320608e-05,
+      "loss": 0.0,
+      "step": 429
+    },
+    {
+      "epoch": 2.356164383561644,
+      "grad_norm": NaN,
+      "learning_rate": 5.182463258982846e-05,
+      "loss": 0.0,
+      "step": 430
+    },
+    {
+      "epoch": 2.361643835616438,
+      "grad_norm": NaN,
+      "learning_rate": 5.127498742746675e-05,
+      "loss": 0.0,
+      "step": 431
+    },
+    {
+      "epoch": 2.367123287671233,
+      "grad_norm": NaN,
+      "learning_rate": 5.072726584517086e-05,
+      "loss": 0.0,
+      "step": 432
+    },
+    {
+      "epoch": 2.3726027397260276,
+      "grad_norm": NaN,
+      "learning_rate": 5.018148946605092e-05,
+      "loss": 0.0,
+      "step": 433
+    },
+    {
+      "epoch": 2.378082191780822,
+      "grad_norm": NaN,
+      "learning_rate": 4.9637679836423924e-05,
+      "loss": 0.0,
+      "step": 434
+    },
+    {
+      "epoch": 2.383561643835616,
+      "grad_norm": NaN,
+      "learning_rate": 4.909585842496287e-05,
+      "loss": 0.0,
+      "step": 435
+    },
+    {
+      "epoch": 2.389041095890411,
+      "grad_norm": NaN,
+      "learning_rate": 4.8556046621849346e-05,
+      "loss": 0.0,
+      "step": 436
+    },
+    {
+      "epoch": 2.3945205479452056,
+      "grad_norm": NaN,
+      "learning_rate": 4.8018265737929044e-05,
+      "loss": 0.0,
+      "step": 437
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": NaN,
+      "learning_rate": 4.748253700387042e-05,
+      "loss": 0.0,
+      "step": 438
+    },
+    {
+      "epoch": 2.4054794520547946,
+      "grad_norm": NaN,
+      "learning_rate": 4.694888156932658e-05,
+      "loss": 0.0,
+      "step": 439
+    },
+    {
+      "epoch": 2.410958904109589,
+      "grad_norm": NaN,
+      "learning_rate": 4.6417320502100316e-05,
+      "loss": 0.0,
+      "step": 440
+    },
+    {
+      "epoch": 2.4164383561643836,
+      "grad_norm": NaN,
+      "learning_rate": 4.588787478731242e-05,
+      "loss": 0.0,
+      "step": 441
+    },
+    {
+      "epoch": 2.421917808219178,
+      "grad_norm": NaN,
+      "learning_rate": 4.5360565326573104e-05,
+      "loss": 0.0,
+      "step": 442
+    },
+    {
+      "epoch": 2.4273972602739726,
+      "grad_norm": NaN,
+      "learning_rate": 4.483541293715698e-05,
+      "loss": 0.0,
+      "step": 443
+    },
+    {
+      "epoch": 2.432876712328767,
+      "grad_norm": NaN,
+      "learning_rate": 4.431243835118124e-05,
+      "loss": 0.0,
+      "step": 444
+    },
+    {
+      "epoch": 2.4383561643835616,
+      "grad_norm": NaN,
+      "learning_rate": 4.379166221478697e-05,
+      "loss": 0.0,
+      "step": 445
+    },
+    {
+      "epoch": 2.4438356164383563,
+      "grad_norm": NaN,
+      "learning_rate": 4.327310508732437e-05,
+      "loss": 0.0,
+      "step": 446
+    },
+    {
+      "epoch": 2.4493150684931506,
+      "grad_norm": NaN,
+      "learning_rate": 4.2756787440540936e-05,
+      "loss": 0.0,
+      "step": 447
+    },
+    {
+      "epoch": 2.4547945205479453,
+      "grad_norm": NaN,
+      "learning_rate": 4.224272965777326e-05,
+      "loss": 0.0,
+      "step": 448
+    },
+    {
+      "epoch": 2.4602739726027396,
+      "grad_norm": NaN,
+      "learning_rate": 4.173095203314241e-05,
+      "loss": 0.0,
+      "step": 449
+    },
+    {
+      "epoch": 2.4657534246575343,
+      "grad_norm": NaN,
+      "learning_rate": 4.12214747707527e-05,
+      "loss": 0.0,
+      "step": 450
+    },
+    {
+      "epoch": 2.4712328767123286,
+      "grad_norm": NaN,
+      "learning_rate": 4.071431798389408e-05,
+      "loss": 0.0,
+      "step": 451
+    },
+    {
+      "epoch": 2.4767123287671233,
+      "grad_norm": NaN,
+      "learning_rate": 4.020950169424815e-05,
+      "loss": 0.0,
+      "step": 452
+    },
+    {
+      "epoch": 2.4821917808219176,
+      "grad_norm": NaN,
+      "learning_rate": 3.9707045831097555e-05,
+      "loss": 0.0,
+      "step": 453
+    },
+    {
+      "epoch": 2.4876712328767123,
+      "grad_norm": NaN,
+      "learning_rate": 3.920697023053949e-05,
+      "loss": 0.0,
+      "step": 454
+    },
+    {
+      "epoch": 2.493150684931507,
+      "grad_norm": NaN,
+      "learning_rate": 3.8709294634702376e-05,
+      "loss": 0.0,
+      "step": 455
+    },
+    {
+      "epoch": 2.4986301369863013,
+      "grad_norm": NaN,
+      "learning_rate": 3.821403869096658e-05,
+      "loss": 0.0,
+      "step": 456
+    },
+    {
+      "epoch": 2.504109589041096,
+      "grad_norm": NaN,
+      "learning_rate": 3.7721221951188765e-05,
+      "loss": 0.0,
+      "step": 457
+    },
+    {
+      "epoch": 2.5095890410958903,
+      "grad_norm": NaN,
+      "learning_rate": 3.7230863870929964e-05,
+      "loss": 0.0,
+      "step": 458
+    },
+    {
+      "epoch": 2.515068493150685,
+      "grad_norm": NaN,
+      "learning_rate": 3.674298380868756e-05,
+      "loss": 0.0,
+      "step": 459
+    },
+    {
+      "epoch": 2.5205479452054793,
+      "grad_norm": NaN,
+      "learning_rate": 3.6257601025131026e-05,
+      "loss": 0.0,
+      "step": 460
+    },
+    {
+      "epoch": 2.526027397260274,
+      "grad_norm": NaN,
+      "learning_rate": 3.577473468234156e-05,
+      "loss": 0.0,
+      "step": 461
+    },
+    {
+      "epoch": 2.5315068493150683,
+      "grad_norm": NaN,
+      "learning_rate": 3.52944038430556e-05,
+      "loss": 0.0,
+      "step": 462
+    },
+    {
+      "epoch": 2.536986301369863,
+      "grad_norm": NaN,
+      "learning_rate": 3.481662746991214e-05,
+      "loss": 0.0,
+      "step": 463
+    },
+    {
+      "epoch": 2.5424657534246577,
+      "grad_norm": NaN,
+      "learning_rate": 3.4341424424704375e-05,
+      "loss": 0.0,
+      "step": 464
+    },
+    {
+      "epoch": 2.547945205479452,
+      "grad_norm": NaN,
+      "learning_rate": 3.386881346763483e-05,
+      "loss": 0.0,
+      "step": 465
+    },
+    {
+      "epoch": 2.5534246575342463,
+      "grad_norm": NaN,
+      "learning_rate": 3.339881325657484e-05,
+      "loss": 0.0,
+      "step": 466
+    },
+    {
+      "epoch": 2.558904109589041,
+      "grad_norm": NaN,
+      "learning_rate": 3.2931442346328004e-05,
+      "loss": 0.0,
+      "step": 467
+    },
+    {
+      "epoch": 2.5643835616438357,
+      "grad_norm": NaN,
+      "learning_rate": 3.246671918789755e-05,
+      "loss": 0.0,
+      "step": 468
+    },
+    {
+      "epoch": 2.56986301369863,
+      "grad_norm": NaN,
+      "learning_rate": 3.200466212775808e-05,
+      "loss": 0.0,
+      "step": 469
+    },
+    {
+      "epoch": 2.5753424657534247,
+      "grad_norm": NaN,
+      "learning_rate": 3.154528940713113e-05,
+      "loss": 0.0,
+      "step": 470
+    },
+    {
+      "epoch": 2.580821917808219,
+      "grad_norm": NaN,
+      "learning_rate": 3.108861916126518e-05,
+      "loss": 0.0,
+      "step": 471
+    },
+    {
+      "epoch": 2.5863013698630137,
+      "grad_norm": NaN,
+      "learning_rate": 3.063466941871952e-05,
+      "loss": 0.0,
+      "step": 472
+    },
+    {
+      "epoch": 2.5917808219178085,
+      "grad_norm": NaN,
+      "learning_rate": 3.018345810065275e-05,
+      "loss": 0.0,
+      "step": 473
+    },
+    {
+      "epoch": 2.5972602739726027,
+      "grad_norm": NaN,
+      "learning_rate": 2.9735003020115092e-05,
+      "loss": 0.0,
+      "step": 474
+    },
+    {
+      "epoch": 2.602739726027397,
+      "grad_norm": NaN,
+      "learning_rate": 2.9289321881345254e-05,
+      "loss": 0.0,
+      "step": 475
+    },
+    {
+      "epoch": 2.6082191780821917,
+      "grad_norm": NaN,
+      "learning_rate": 2.8846432279071467e-05,
+      "loss": 0.0,
+      "step": 476
+    },
+    {
+      "epoch": 2.6136986301369864,
+      "grad_norm": NaN,
+      "learning_rate": 2.840635169781688e-05,
+      "loss": 0.0,
+      "step": 477
+    },
+    {
+      "epoch": 2.6191780821917807,
+      "grad_norm": NaN,
+      "learning_rate": 2.7969097511209308e-05,
+      "loss": 0.0,
+      "step": 478
+    },
+    {
+      "epoch": 2.6246575342465754,
+      "grad_norm": NaN,
+      "learning_rate": 2.753468698129533e-05,
+      "loss": 0.0,
+      "step": 479
+    },
+    {
+      "epoch": 2.6301369863013697,
+      "grad_norm": NaN,
+      "learning_rate": 2.7103137257858868e-05,
+      "loss": 0.0,
+      "step": 480
+    },
+    {
+      "epoch": 2.6356164383561644,
+      "grad_norm": NaN,
+      "learning_rate": 2.6674465377744017e-05,
+      "loss": 0.0,
+      "step": 481
+    },
+    {
+      "epoch": 2.641095890410959,
+      "grad_norm": NaN,
+      "learning_rate": 2.624868826418262e-05,
+      "loss": 0.0,
+      "step": 482
+    },
+    {
+      "epoch": 2.6465753424657534,
+      "grad_norm": NaN,
+      "learning_rate": 2.582582272612609e-05,
+      "loss": 0.0,
+      "step": 483
+    },
+    {
+      "epoch": 2.6520547945205477,
+      "grad_norm": NaN,
+      "learning_rate": 2.540588545758179e-05,
+      "loss": 0.0,
+      "step": 484
+    },
+    {
+      "epoch": 2.6575342465753424,
+      "grad_norm": NaN,
+      "learning_rate": 2.4988893036954043e-05,
+      "loss": 0.0,
+      "step": 485
+    },
+    {
+      "epoch": 2.663013698630137,
+      "grad_norm": NaN,
+      "learning_rate": 2.4574861926389615e-05,
+      "loss": 0.0,
+      "step": 486
+    },
+    {
+      "epoch": 2.6684931506849314,
+      "grad_norm": NaN,
+      "learning_rate": 2.4163808471127812e-05,
+      "loss": 0.0,
+      "step": 487
+    },
+    {
+      "epoch": 2.673972602739726,
+      "grad_norm": NaN,
+      "learning_rate": 2.37557488988552e-05,
+      "loss": 0.0,
+      "step": 488
+    },
+    {
+      "epoch": 2.6794520547945204,
+      "grad_norm": NaN,
+      "learning_rate": 2.3350699319065026e-05,
+      "loss": 0.0,
+      "step": 489
+    },
+    {
+      "epoch": 2.684931506849315,
+      "grad_norm": NaN,
+      "learning_rate": 2.2948675722421086e-05,
+      "loss": 0.0,
+      "step": 490
+    },
+    {
+      "epoch": 2.69041095890411,
+      "grad_norm": NaN,
+      "learning_rate": 2.254969398012663e-05,
+      "loss": 0.0,
+      "step": 491
+    },
+    {
+      "epoch": 2.695890410958904,
+      "grad_norm": NaN,
+      "learning_rate": 2.2153769843297667e-05,
+      "loss": 0.0,
+      "step": 492
+    },
+    {
+      "epoch": 2.7013698630136984,
+      "grad_norm": NaN,
+      "learning_rate": 2.1760918942341192e-05,
+      "loss": 0.0,
+      "step": 493
+    },
+    {
+      "epoch": 2.706849315068493,
+      "grad_norm": NaN,
+      "learning_rate": 2.137115678633811e-05,
+      "loss": 0.0,
+      "step": 494
+    },
+    {
+      "epoch": 2.712328767123288,
+      "grad_norm": NaN,
+      "learning_rate": 2.098449876243096e-05,
+      "loss": 0.0,
+      "step": 495
+    },
+    {
+      "epoch": 2.717808219178082,
+      "grad_norm": NaN,
+      "learning_rate": 2.0600960135216462e-05,
+      "loss": 0.0,
+      "step": 496
+    },
+    {
+      "epoch": 2.723287671232877,
+      "grad_norm": NaN,
+      "learning_rate": 2.0220556046142893e-05,
+      "loss": 0.0,
+      "step": 497
+    },
+    {
+      "epoch": 2.728767123287671,
+      "grad_norm": NaN,
+      "learning_rate": 1.9843301512912327e-05,
+      "loss": 0.0,
+      "step": 498
+    },
+    {
+      "epoch": 2.734246575342466,
+      "grad_norm": NaN,
+      "learning_rate": 1.946921142888781e-05,
+      "loss": 0.0,
+      "step": 499
+    },
+    {
+      "epoch": 2.73972602739726,
+      "grad_norm": NaN,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 0.0,
+      "step": 500
+    },
+    {
+      "epoch": 2.745205479452055,
+      "grad_norm": NaN,
+      "learning_rate": 1.8730583556690605e-05,
+      "loss": 0.0,
+      "step": 501
+    },
+    {
+      "epoch": 2.750684931506849,
+      "grad_norm": NaN,
+      "learning_rate": 1.8366074928281607e-05,
+      "loss": 0.0,
+      "step": 502
+    },
+    {
+      "epoch": 2.756164383561644,
+      "grad_norm": NaN,
+      "learning_rate": 1.8004789067454764e-05,
+      "loss": 0.0,
+      "step": 503
+    },
+    {
+      "epoch": 2.7616438356164386,
+      "grad_norm": NaN,
+      "learning_rate": 1.7646740237157256e-05,
+      "loss": 0.0,
+      "step": 504
+    },
+    {
+      "epoch": 2.767123287671233,
+      "grad_norm": NaN,
+      "learning_rate": 1.7291942572543807e-05,
+      "loss": 0.0,
+      "step": 505
+    },
+    {
+      "epoch": 2.7726027397260276,
+      "grad_norm": NaN,
+      "learning_rate": 1.6940410080418723e-05,
+      "loss": 0.0,
+      "step": 506
+    },
+    {
+      "epoch": 2.778082191780822,
+      "grad_norm": NaN,
+      "learning_rate": 1.6592156638682886e-05,
+      "loss": 0.0,
+      "step": 507
+    },
+    {
+      "epoch": 2.7835616438356166,
+      "grad_norm": NaN,
+      "learning_rate": 1.6247195995785837e-05,
+      "loss": 0.0,
+      "step": 508
+    },
+    {
+      "epoch": 2.789041095890411,
+      "grad_norm": NaN,
+      "learning_rate": 1.5905541770183096e-05,
+      "loss": 0.0,
+      "step": 509
+    },
+    {
+      "epoch": 2.7945205479452055,
+      "grad_norm": NaN,
+      "learning_rate": 1.5567207449798515e-05,
+      "loss": 0.0,
+      "step": 510
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": NaN,
+      "learning_rate": 1.5232206391491699e-05,
+      "loss": 0.0,
+      "step": 511
+    },
+    {
+      "epoch": 2.8054794520547945,
+      "grad_norm": NaN,
+      "learning_rate": 1.4900551820530828e-05,
+      "loss": 0.0,
+      "step": 512
+    },
+    {
+      "epoch": 2.8109589041095893,
+      "grad_norm": NaN,
+      "learning_rate": 1.4572256830070497e-05,
+      "loss": 0.0,
+      "step": 513
+    },
+    {
+      "epoch": 2.8164383561643835,
+      "grad_norm": NaN,
+      "learning_rate": 1.4247334380634792e-05,
+      "loss": 0.0,
+      "step": 514
+    },
+    {
+      "epoch": 2.821917808219178,
+      "grad_norm": NaN,
+      "learning_rate": 1.3925797299605647e-05,
+      "loss": 0.0,
+      "step": 515
+    },
+    {
+      "epoch": 2.8273972602739725,
+      "grad_norm": NaN,
+      "learning_rate": 1.3607658280716473e-05,
+      "loss": 0.0,
+      "step": 516
+    },
+    {
+      "epoch": 2.8328767123287673,
+      "grad_norm": NaN,
+      "learning_rate": 1.3292929883550998e-05,
+      "loss": 0.0,
+      "step": 517
+    },
+    {
+      "epoch": 2.8383561643835615,
+      "grad_norm": NaN,
+      "learning_rate": 1.2981624533047432e-05,
+      "loss": 0.0,
+      "step": 518
+    },
+    {
+      "epoch": 2.8438356164383563,
+      "grad_norm": NaN,
+      "learning_rate": 1.2673754519008008e-05,
+      "loss": 0.0,
+      "step": 519
+    },
+    {
+      "epoch": 2.8493150684931505,
+      "grad_norm": NaN,
+      "learning_rate": 1.2369331995613665e-05,
+      "loss": 0.0,
+      "step": 520
+    },
+    {
+      "epoch": 2.8547945205479452,
+      "grad_norm": NaN,
+      "learning_rate": 1.206836898094439e-05,
+      "loss": 0.0,
+      "step": 521
+    },
+    {
+      "epoch": 2.86027397260274,
+      "grad_norm": NaN,
+      "learning_rate": 1.1770877356504683e-05,
+      "loss": 0.0,
+      "step": 522
+    },
+    {
+      "epoch": 2.8657534246575342,
+      "grad_norm": NaN,
+      "learning_rate": 1.1476868866754486e-05,
+      "loss": 0.0,
+      "step": 523
+    },
+    {
+      "epoch": 2.8712328767123285,
+      "grad_norm": NaN,
+      "learning_rate": 1.1186355118645554e-05,
+      "loss": 0.0,
+      "step": 524
+    },
+    {
+      "epoch": 2.8767123287671232,
+      "grad_norm": NaN,
+      "learning_rate": 1.0899347581163221e-05,
+      "loss": 0.0,
+      "step": 525
+    },
+    {
+      "epoch": 2.882191780821918,
+      "grad_norm": NaN,
+      "learning_rate": 1.0615857584873623e-05,
+      "loss": 0.0,
+      "step": 526
+    },
+    {
+      "epoch": 2.8876712328767122,
+      "grad_norm": NaN,
+      "learning_rate": 1.0335896321476413e-05,
+      "loss": 0.0,
+      "step": 527
+    },
+    {
+      "epoch": 2.893150684931507,
+      "grad_norm": NaN,
+      "learning_rate": 1.0059474843362892e-05,
+      "loss": 0.0,
+      "step": 528
+    },
+    {
+      "epoch": 2.8986301369863012,
+      "grad_norm": NaN,
+      "learning_rate": 9.786604063179728e-06,
+      "loss": 0.0,
+      "step": 529
+    },
+    {
+      "epoch": 2.904109589041096,
+      "grad_norm": NaN,
+      "learning_rate": 9.517294753398064e-06,
+      "loss": 0.0,
+      "step": 530
+    },
+    {
+      "epoch": 2.9095890410958907,
+      "grad_norm": NaN,
+      "learning_rate": 9.251557545888312e-06,
+      "loss": 0.0,
+      "step": 531
+    },
+    {
+      "epoch": 2.915068493150685,
+      "grad_norm": NaN,
+      "learning_rate": 8.989402931500434e-06,
+      "loss": 0.0,
+      "step": 532
+    },
+    {
+      "epoch": 2.9205479452054792,
+      "grad_norm": NaN,
+      "learning_rate": 8.730841259649725e-06,
+      "loss": 0.0,
+      "step": 533
+    },
+    {
+      "epoch": 2.926027397260274,
+      "grad_norm": NaN,
+      "learning_rate": 8.475882737908248e-06,
+      "loss": 0.0,
+      "step": 534
+    },
+    {
+      "epoch": 2.9315068493150687,
+      "grad_norm": NaN,
+      "learning_rate": 8.224537431601886e-06,
+      "loss": 0.0,
+      "step": 535
+    },
+    {
+      "epoch": 2.936986301369863,
+      "grad_norm": NaN,
+      "learning_rate": 7.976815263412963e-06,
+      "loss": 0.0,
+      "step": 536
+    },
+    {
+      "epoch": 2.9424657534246577,
+      "grad_norm": NaN,
+      "learning_rate": 7.73272601298851e-06,
+      "loss": 0.0,
+      "step": 537
+    },
+    {
+      "epoch": 2.947945205479452,
+      "grad_norm": NaN,
+      "learning_rate": 7.492279316554207e-06,
+      "loss": 0.0,
+      "step": 538
+    },
+    {
+      "epoch": 2.9534246575342467,
+      "grad_norm": NaN,
+      "learning_rate": 7.255484666533874e-06,
+      "loss": 0.0,
+      "step": 539
+    },
+    {
+      "epoch": 2.958904109589041,
+      "grad_norm": NaN,
+      "learning_rate": 7.022351411174866e-06,
+      "loss": 0.0,
+      "step": 540
+    },
+    {
+      "epoch": 2.9643835616438357,
+      "grad_norm": NaN,
+      "learning_rate": 6.7928887541789055e-06,
+      "loss": 0.0,
+      "step": 541
+    },
+    {
+      "epoch": 2.96986301369863,
+      "grad_norm": NaN,
+      "learning_rate": 6.5671057543387985e-06,
+      "loss": 0.0,
+      "step": 542
+    },
+    {
+      "epoch": 2.9753424657534246,
+      "grad_norm": NaN,
+      "learning_rate": 6.345011325180772e-06,
+      "loss": 0.0,
+      "step": 543
+    },
+    {
+      "epoch": 2.9808219178082194,
+      "grad_norm": NaN,
+      "learning_rate": 6.126614234612593e-06,
+      "loss": 0.0,
+      "step": 544
+    },
+    {
+      "epoch": 2.9863013698630136,
+      "grad_norm": NaN,
+      "learning_rate": 5.911923104577455e-06,
+      "loss": 0.0,
+      "step": 545
+    },
+    {
+      "epoch": 2.9917808219178084,
+      "grad_norm": NaN,
+      "learning_rate": 5.700946410713548e-06,
+      "loss": 0.0,
+      "step": 546
+    },
+    {
+      "epoch": 2.9972602739726026,
+      "grad_norm": NaN,
+      "learning_rate": 5.49369248201953e-06,
+      "loss": 0.0,
+      "step": 547
+    },
+    {
+      "epoch": 2.9972602739726026,
+      "eval_loss": NaN,
+      "eval_runtime": 18.511,
+      "eval_samples_per_second": 66.393,
+      "eval_steps_per_second": 2.107,
+      "step": 547
     }
   ],
   "logging_steps": 1,
@@ -2597,7 +3879,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 9.215405468984279e+17,
+  "total_flos": 1.3824862435892265e+18,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null