Training in progress, step 20500, checkpoint

checkpoint-20500/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:605affbbd497288577a43d690c8ff36873c9670173c561a762c5cc9f46164ba5
+oid sha256:85ae5aac4408458f02fc50e38541d0df7ca94d3b23ab092be694e2cfe6f6fa9b
 size 1711848436

checkpoint-20500/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ed1270fa2f64c09fb28c03edad6d9115c7e077500d4eecdae97d0c4fbf3e07eb
+oid sha256:95d6d7eb66b461ec4f1acbbdd55eedcf76de0a740850abd7541489711f857b78
 size 3424043887

checkpoint-20500/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:551bd9fefe21a0210b8f009f7e379124ea37132a2e134634bfee8fc680d70fd3
+oid sha256:0112b25beedbe5c58a79d56a364c54bf302957f0afcdbd9d18fb998c348fc8c2
 size 14503

checkpoint-20500/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b208ec9f61a3a4f407c35167568a6fe876bd6f6057200867c82e458fe6281eda
+oid sha256:07b6db9b93ae41ac6ac4614304492a9d795c09aed3a76f855e00419eb720a960
 size 623

checkpoint-20500/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.7746543884277344,
+  "best_metric": 0.8584741353988647,
   "best_model_checkpoint": "./output/clip-finetuned-csu-p14-336-e4l58-l/checkpoint-20500",
-  "epoch": 3.776713338246131,
+  "epoch": 2.186433447098976,
   "eval_steps": 500,
   "global_step": 20500,
   "is_hyper_param_search": false,
@@ -9,623 +9,623 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.09209799226376865,
-      "grad_norm": 0.0014697719598188996,
-      "learning_rate": 4.884877509670289e-08,
-      "loss": 0.4588,
+      "epoch": 0.05332764505119454,
+      "grad_norm": 217.55360412597656,
+      "learning_rate": 4.933340443686007e-08,
+      "loss": 0.4667,
       "step": 500
     },
     {
-      "epoch": 0.09209799226376865,
-      "eval_loss": 1.4194883108139038,
-      "eval_runtime": 78.7703,
-      "eval_samples_per_second": 15.31,
-      "eval_steps_per_second": 1.917,
+      "epoch": 0.05332764505119454,
+      "eval_loss": 1.4426143169403076,
+      "eval_runtime": 61.9901,
+      "eval_samples_per_second": 15.922,
+      "eval_steps_per_second": 2.0,
       "step": 500
     },
     {
-      "epoch": 0.1841959845275373,
-      "grad_norm": 301.76043701171875,
-      "learning_rate": 4.769755019340578e-08,
-      "loss": 0.4255,
+      "epoch": 0.10665529010238908,
+      "grad_norm": 73.73340606689453,
+      "learning_rate": 4.8666808873720136e-08,
+      "loss": 0.4532,
       "step": 1000
     },
     {
-      "epoch": 0.1841959845275373,
-      "eval_loss": 1.3416553735733032,
-      "eval_runtime": 77.0051,
-      "eval_samples_per_second": 15.661,
-      "eval_steps_per_second": 1.961,
+      "epoch": 0.10665529010238908,
+      "eval_loss": 1.3815597295761108,
+      "eval_runtime": 62.4624,
+      "eval_samples_per_second": 15.801,
+      "eval_steps_per_second": 1.985,
       "step": 1000
     },
     {
-      "epoch": 0.27629397679130596,
-      "grad_norm": 35.28853225708008,
-      "learning_rate": 4.654632529010867e-08,
-      "loss": 0.3724,
+      "epoch": 0.1599829351535836,
+      "grad_norm": 471.4011535644531,
+      "learning_rate": 4.80002133105802e-08,
+      "loss": 0.3749,
       "step": 1500
     },
     {
-      "epoch": 0.27629397679130596,
-      "eval_loss": 1.2872973680496216,
-      "eval_runtime": 77.5439,
-      "eval_samples_per_second": 15.552,
-      "eval_steps_per_second": 1.947,
+      "epoch": 0.1599829351535836,
+      "eval_loss": 1.3310539722442627,
+      "eval_runtime": 63.3918,
+      "eval_samples_per_second": 15.57,
+      "eval_steps_per_second": 1.956,
       "step": 1500
     },
     {
-      "epoch": 0.3683919690550746,
-      "grad_norm": 338.9349060058594,
-      "learning_rate": 4.539510038681156e-08,
-      "loss": 0.3251,
+      "epoch": 0.21331058020477817,
+      "grad_norm": 11.962693214416504,
+      "learning_rate": 4.733361774744027e-08,
+      "loss": 0.336,
       "step": 2000
     },
     {
-      "epoch": 0.3683919690550746,
-      "eval_loss": 1.234910249710083,
-      "eval_runtime": 78.7715,
-      "eval_samples_per_second": 15.31,
-      "eval_steps_per_second": 1.917,
+      "epoch": 0.21331058020477817,
+      "eval_loss": 1.2890639305114746,
+      "eval_runtime": 63.177,
+      "eval_samples_per_second": 15.623,
+      "eval_steps_per_second": 1.963,
       "step": 2000
     },
     {
-      "epoch": 0.46048996131884323,
-      "grad_norm": 325.85101318359375,
-      "learning_rate": 4.4243875483514457e-08,
-      "loss": 0.3308,
+      "epoch": 0.2666382252559727,
+      "grad_norm": 375.1609191894531,
+      "learning_rate": 4.666702218430034e-08,
+      "loss": 0.3585,
       "step": 2500
     },
     {
-      "epoch": 0.46048996131884323,
-      "eval_loss": 1.1944907903671265,
-      "eval_runtime": 79.0397,
-      "eval_samples_per_second": 15.258,
-      "eval_steps_per_second": 1.91,
+      "epoch": 0.2666382252559727,
+      "eval_loss": 1.2536433935165405,
+      "eval_runtime": 63.1512,
+      "eval_samples_per_second": 15.629,
+      "eval_steps_per_second": 1.964,
       "step": 2500
     },
     {
-      "epoch": 0.5525879535826119,
-      "grad_norm": 100.85855102539062,
-      "learning_rate": 4.309265058021735e-08,
-      "loss": 0.3017,
+      "epoch": 0.3199658703071672,
+      "grad_norm": 0.0022936267778277397,
+      "learning_rate": 4.600042662116041e-08,
+      "loss": 0.303,
       "step": 3000
     },
     {
-      "epoch": 0.5525879535826119,
-      "eval_loss": 1.1593009233474731,
-      "eval_runtime": 77.6354,
-      "eval_samples_per_second": 15.534,
-      "eval_steps_per_second": 1.945,
+      "epoch": 0.3199658703071672,
+      "eval_loss": 1.2202869653701782,
+      "eval_runtime": 63.5857,
+      "eval_samples_per_second": 15.522,
+      "eval_steps_per_second": 1.95,
       "step": 3000
     },
     {
-      "epoch": 0.6446859458463805,
-      "grad_norm": 8.865598678588867,
-      "learning_rate": 4.194142567692024e-08,
-      "loss": 0.2962,
+      "epoch": 0.37329351535836175,
+      "grad_norm": 15.805087089538574,
+      "learning_rate": 4.5333831058020476e-08,
+      "loss": 0.3242,
       "step": 3500
     },
     {
-      "epoch": 0.6446859458463805,
-      "eval_loss": 1.1259427070617676,
-      "eval_runtime": 78.2715,
-      "eval_samples_per_second": 15.408,
-      "eval_steps_per_second": 1.929,
+      "epoch": 0.37329351535836175,
+      "eval_loss": 1.195627212524414,
+      "eval_runtime": 63.4024,
+      "eval_samples_per_second": 15.567,
+      "eval_steps_per_second": 1.956,
       "step": 3500
     },
     {
-      "epoch": 0.7367839381101492,
-      "grad_norm": 4.393447399139404,
-      "learning_rate": 4.079020077362314e-08,
-      "loss": 0.2919,
+      "epoch": 0.42662116040955633,
+      "grad_norm": 5.377908229827881,
+      "learning_rate": 4.4667235494880546e-08,
+      "loss": 0.2427,
       "step": 4000
     },
     {
-      "epoch": 0.7367839381101492,
-      "eval_loss": 1.0954023599624634,
-      "eval_runtime": 78.5366,
-      "eval_samples_per_second": 15.356,
-      "eval_steps_per_second": 1.923,
+      "epoch": 0.42662116040955633,
+      "eval_loss": 1.169384241104126,
+      "eval_runtime": 63.5082,
+      "eval_samples_per_second": 15.541,
+      "eval_steps_per_second": 1.953,
       "step": 4000
     },
     {
-      "epoch": 0.8288819303739179,
-      "grad_norm": 1.6036031246185303,
-      "learning_rate": 3.9638975870326026e-08,
-      "loss": 0.307,
+      "epoch": 0.47994880546075086,
+      "grad_norm": 150.5334930419922,
+      "learning_rate": 4.4000639931740615e-08,
+      "loss": 0.2993,
       "step": 4500
     },
     {
-      "epoch": 0.8288819303739179,
-      "eval_loss": 1.0728861093521118,
-      "eval_runtime": 78.4032,
-      "eval_samples_per_second": 15.382,
-      "eval_steps_per_second": 1.926,
+      "epoch": 0.47994880546075086,
+      "eval_loss": 1.145558476448059,
+      "eval_runtime": 62.3888,
+      "eval_samples_per_second": 15.82,
+      "eval_steps_per_second": 1.988,
       "step": 4500
     },
     {
-      "epoch": 0.9209799226376865,
-      "grad_norm": 11.990758895874023,
-      "learning_rate": 3.8487750967028915e-08,
-      "loss": 0.2764,
+      "epoch": 0.5332764505119454,
+      "grad_norm": 0.12246542423963547,
+      "learning_rate": 4.333404436860068e-08,
+      "loss": 0.3183,
       "step": 5000
     },
     {
-      "epoch": 0.9209799226376865,
-      "eval_loss": 1.0523593425750732,
-      "eval_runtime": 78.3726,
-      "eval_samples_per_second": 15.388,
-      "eval_steps_per_second": 1.927,
+      "epoch": 0.5332764505119454,
+      "eval_loss": 1.1201218366622925,
+      "eval_runtime": 63.7154,
+      "eval_samples_per_second": 15.491,
+      "eval_steps_per_second": 1.946,
       "step": 5000
     },
     {
-      "epoch": 1.013077914901455,
-      "grad_norm": 43.865562438964844,
-      "learning_rate": 3.733652606373181e-08,
-      "loss": 0.2456,
+      "epoch": 0.58660409556314,
+      "grad_norm": 0.3306196630001068,
+      "learning_rate": 4.266744880546075e-08,
+      "loss": 0.307,
       "step": 5500
     },
     {
-      "epoch": 1.013077914901455,
-      "eval_loss": 1.0375442504882812,
-      "eval_runtime": 78.2473,
-      "eval_samples_per_second": 15.413,
-      "eval_steps_per_second": 1.93,
+      "epoch": 0.58660409556314,
+      "eval_loss": 1.0982334613800049,
+      "eval_runtime": 62.4882,
+      "eval_samples_per_second": 15.795,
+      "eval_steps_per_second": 1.984,
       "step": 5500
     },
     {
-      "epoch": 1.1051759071652238,
-      "grad_norm": 105.91036987304688,
-      "learning_rate": 3.61853011604347e-08,
-      "loss": 0.2642,
+      "epoch": 0.6399317406143344,
+      "grad_norm": 149.38189697265625,
+      "learning_rate": 4.2000853242320816e-08,
+      "loss": 0.2638,
       "step": 6000
     },
     {
-      "epoch": 1.1051759071652238,
-      "eval_loss": 1.0233356952667236,
-      "eval_runtime": 78.3135,
-      "eval_samples_per_second": 15.4,
-      "eval_steps_per_second": 1.928,
+      "epoch": 0.6399317406143344,
+      "eval_loss": 1.0779541730880737,
+      "eval_runtime": 63.4101,
+      "eval_samples_per_second": 15.565,
+      "eval_steps_per_second": 1.956,
       "step": 6000
     },
     {
-      "epoch": 1.1972738994289924,
-      "grad_norm": 221.4896240234375,
-      "learning_rate": 3.503407625713759e-08,
-      "loss": 0.2066,
+      "epoch": 0.693259385665529,
+      "grad_norm": 353.0036315917969,
+      "learning_rate": 4.1334257679180886e-08,
+      "loss": 0.2226,
       "step": 6500
     },
     {
-      "epoch": 1.1972738994289924,
-      "eval_loss": 1.0104238986968994,
-      "eval_runtime": 78.6485,
-      "eval_samples_per_second": 15.334,
-      "eval_steps_per_second": 1.92,
+      "epoch": 0.693259385665529,
+      "eval_loss": 1.0613036155700684,
+      "eval_runtime": 62.8292,
+      "eval_samples_per_second": 15.709,
+      "eval_steps_per_second": 1.974,
       "step": 6500
     },
     {
-      "epoch": 1.289371891692761,
-      "grad_norm": 270.53253173828125,
-      "learning_rate": 3.3882851353840485e-08,
-      "loss": 0.2376,
+      "epoch": 0.7465870307167235,
+      "grad_norm": 0.13188917934894562,
+      "learning_rate": 4.0667662116040955e-08,
+      "loss": 0.2453,
       "step": 7000
     },
     {
-      "epoch": 1.289371891692761,
-      "eval_loss": 0.998353898525238,
-      "eval_runtime": 78.5502,
-      "eval_samples_per_second": 15.353,
-      "eval_steps_per_second": 1.922,
+      "epoch": 0.7465870307167235,
+      "eval_loss": 1.0443507432937622,
+      "eval_runtime": 62.7457,
+      "eval_samples_per_second": 15.73,
+      "eval_steps_per_second": 1.976,
       "step": 7000
     },
     {
-      "epoch": 1.3814698839565298,
-      "grad_norm": 402.9339904785156,
-      "learning_rate": 3.2731626450543374e-08,
-      "loss": 0.1931,
+      "epoch": 0.7999146757679181,
+      "grad_norm": 59.391639709472656,
+      "learning_rate": 4.000106655290102e-08,
+      "loss": 0.272,
       "step": 7500
     },
     {
-      "epoch": 1.3814698839565298,
-      "eval_loss": 0.9887453317642212,
-      "eval_runtime": 78.4833,
-      "eval_samples_per_second": 15.366,
-      "eval_steps_per_second": 1.924,
+      "epoch": 0.7999146757679181,
+      "eval_loss": 1.0300519466400146,
+      "eval_runtime": 62.3719,
+      "eval_samples_per_second": 15.824,
+      "eval_steps_per_second": 1.988,
       "step": 7500
     },
     {
-      "epoch": 1.4735678762202984,
-      "grad_norm": 0.006355441175401211,
-      "learning_rate": 3.158040154724626e-08,
-      "loss": 0.2163,
+      "epoch": 0.8532423208191127,
+      "grad_norm": 495.4797668457031,
+      "learning_rate": 3.9334470989761093e-08,
+      "loss": 0.283,
       "step": 8000
     },
     {
-      "epoch": 1.4735678762202984,
-      "eval_loss": 0.9767189621925354,
-      "eval_runtime": 78.5499,
-      "eval_samples_per_second": 15.353,
-      "eval_steps_per_second": 1.922,
+      "epoch": 0.8532423208191127,
+      "eval_loss": 1.0166898965835571,
+      "eval_runtime": 62.0338,
+      "eval_samples_per_second": 15.911,
+      "eval_steps_per_second": 1.999,
       "step": 8000
     },
     {
-      "epoch": 1.565665868484067,
-      "grad_norm": 0.12982100248336792,
-      "learning_rate": 3.042917664394916e-08,
-      "loss": 0.1903,
+      "epoch": 0.9065699658703071,
+      "grad_norm": 159.0310821533203,
+      "learning_rate": 3.866787542662116e-08,
+      "loss": 0.2331,
       "step": 8500
     },
     {
-      "epoch": 1.565665868484067,
-      "eval_loss": 0.9664921760559082,
-      "eval_runtime": 78.6832,
-      "eval_samples_per_second": 15.327,
-      "eval_steps_per_second": 1.919,
+      "epoch": 0.9065699658703071,
+      "eval_loss": 1.003544807434082,
+      "eval_runtime": 62.9704,
+      "eval_samples_per_second": 15.674,
+      "eval_steps_per_second": 1.969,
       "step": 8500
     },
     {
-      "epoch": 1.6577638607478358,
-      "grad_norm": 173.31788635253906,
-      "learning_rate": 2.927795174065205e-08,
-      "loss": 0.2069,
+      "epoch": 0.9598976109215017,
+      "grad_norm": 45.40327835083008,
+      "learning_rate": 3.8001279863481225e-08,
+      "loss": 0.2362,
       "step": 9000
     },
     {
-      "epoch": 1.6577638607478358,
-      "eval_loss": 0.9571623206138611,
-      "eval_runtime": 78.544,
-      "eval_samples_per_second": 15.354,
-      "eval_steps_per_second": 1.922,
+      "epoch": 0.9598976109215017,
+      "eval_loss": 0.992453396320343,
+      "eval_runtime": 63.1405,
+      "eval_samples_per_second": 15.632,
+      "eval_steps_per_second": 1.964,
       "step": 9000
     },
     {
-      "epoch": 1.7498618530116044,
-      "grad_norm": 358.7403259277344,
-      "learning_rate": 2.8126726837354947e-08,
-      "loss": 0.2093,
+      "epoch": 1.0132252559726962,
+      "grad_norm": 621.8538818359375,
+      "learning_rate": 3.7334684300341295e-08,
+      "loss": 0.2396,
       "step": 9500
     },
     {
-      "epoch": 1.7498618530116044,
-      "eval_loss": 0.9496769905090332,
-      "eval_runtime": 78.3124,
-      "eval_samples_per_second": 15.4,
-      "eval_steps_per_second": 1.928,
+      "epoch": 1.0132252559726962,
+      "eval_loss": 0.9829725027084351,
+      "eval_runtime": 62.3244,
+      "eval_samples_per_second": 15.836,
+      "eval_steps_per_second": 1.99,
       "step": 9500
     },
     {
-      "epoch": 1.841959845275373,
-      "grad_norm": 435.9449768066406,
-      "learning_rate": 2.6975501934057836e-08,
-      "loss": 0.2523,
+      "epoch": 1.0665529010238908,
+      "grad_norm": 461.60736083984375,
+      "learning_rate": 3.6668088737201364e-08,
+      "loss": 0.2013,
       "step": 10000
     },
     {
-      "epoch": 1.841959845275373,
-      "eval_loss": 0.9420022368431091,
-      "eval_runtime": 78.3702,
-      "eval_samples_per_second": 15.388,
-      "eval_steps_per_second": 1.927,
+      "epoch": 1.0665529010238908,
+      "eval_loss": 0.9735796451568604,
+      "eval_runtime": 63.0763,
+      "eval_samples_per_second": 15.648,
+      "eval_steps_per_second": 1.966,
       "step": 10000
     },
     {
-      "epoch": 1.9340578375391417,
-      "grad_norm": 23.49118423461914,
-      "learning_rate": 2.582427703076073e-08,
-      "loss": 0.2127,
+      "epoch": 1.1198805460750854,
+      "grad_norm": 7.0309906005859375,
+      "learning_rate": 3.6001493174061433e-08,
+      "loss": 0.2082,
       "step": 10500
     },
     {
-      "epoch": 1.9340578375391417,
-      "eval_loss": 0.9328628182411194,
-      "eval_runtime": 76.6904,
-      "eval_samples_per_second": 15.726,
-      "eval_steps_per_second": 1.969,
+      "epoch": 1.1198805460750854,
+      "eval_loss": 0.9639086127281189,
+      "eval_runtime": 63.0238,
+      "eval_samples_per_second": 15.661,
+      "eval_steps_per_second": 1.968,
       "step": 10500
     },
     {
-      "epoch": 2.02615582980291,
-      "grad_norm": 0.4029064476490021,
-      "learning_rate": 2.467305212746362e-08,
-      "loss": 0.1968,
+      "epoch": 1.17320819112628,
+      "grad_norm": 304.01995849609375,
+      "learning_rate": 3.5334897610921496e-08,
+      "loss": 0.2023,
       "step": 11000
     },
     {
-      "epoch": 2.02615582980291,
-      "eval_loss": 0.926975429058075,
-      "eval_runtime": 76.7161,
-      "eval_samples_per_second": 15.72,
-      "eval_steps_per_second": 1.968,
+      "epoch": 1.17320819112628,
+      "eval_loss": 0.9558469653129578,
+      "eval_runtime": 62.1866,
+      "eval_samples_per_second": 15.872,
+      "eval_steps_per_second": 1.994,
       "step": 11000
     },
     {
-      "epoch": 2.118253822066679,
-      "grad_norm": 274.0943603515625,
-      "learning_rate": 2.3521827224166513e-08,
-      "loss": 0.1879,
+      "epoch": 1.2265358361774745,
+      "grad_norm": 0.0017981340643018484,
+      "learning_rate": 3.4668302047781565e-08,
+      "loss": 0.2331,
       "step": 11500
     },
     {
-      "epoch": 2.118253822066679,
-      "eval_loss": 0.9231382012367249,
-      "eval_runtime": 78.6224,
-      "eval_samples_per_second": 15.339,
-      "eval_steps_per_second": 1.921,
+      "epoch": 1.2265358361774745,
+      "eval_loss": 0.9465070962905884,
+      "eval_runtime": 63.1547,
+      "eval_samples_per_second": 15.628,
+      "eval_steps_per_second": 1.963,
       "step": 11500
     },
     {
-      "epoch": 2.2103518143304477,
-      "grad_norm": 0.000765918695833534,
-      "learning_rate": 2.2370602320869402e-08,
-      "loss": 0.1981,
+      "epoch": 1.2798634812286689,
+      "grad_norm": 0.8147886395454407,
+      "learning_rate": 3.4001706484641635e-08,
+      "loss": 0.1784,
       "step": 12000
     },
     {
-      "epoch": 2.2103518143304477,
-      "eval_loss": 0.9183884263038635,
-      "eval_runtime": 76.7707,
-      "eval_samples_per_second": 15.709,
-      "eval_steps_per_second": 1.967,
+      "epoch": 1.2798634812286689,
+      "eval_loss": 0.9392343163490295,
+      "eval_runtime": 62.9221,
+      "eval_samples_per_second": 15.686,
+      "eval_steps_per_second": 1.971,
       "step": 12000
     },
     {
-      "epoch": 2.302449806594216,
-      "grad_norm": 0.6250036358833313,
-      "learning_rate": 2.1219377417572295e-08,
-      "loss": 0.1964,
+      "epoch": 1.3331911262798635,
+      "grad_norm": 0.01276802271604538,
+      "learning_rate": 3.3335110921501704e-08,
+      "loss": 0.1953,
       "step": 12500
     },
     {
-      "epoch": 2.302449806594216,
-      "eval_loss": 0.9134928584098816,
-      "eval_runtime": 78.5128,
-      "eval_samples_per_second": 15.361,
-      "eval_steps_per_second": 1.923,
+      "epoch": 1.3331911262798635,
+      "eval_loss": 0.9316345453262329,
+      "eval_runtime": 62.2816,
+      "eval_samples_per_second": 15.847,
+      "eval_steps_per_second": 1.991,
       "step": 12500
     },
     {
-      "epoch": 2.394547798857985,
-      "grad_norm": 158.6990203857422,
-      "learning_rate": 2.0068152514275187e-08,
-      "loss": 0.1697,
+      "epoch": 1.386518771331058,
+      "grad_norm": 0.07341516762971878,
+      "learning_rate": 3.2668515358361773e-08,
+      "loss": 0.1867,
       "step": 13000
     },
     {
-      "epoch": 2.394547798857985,
-      "eval_loss": 0.910048246383667,
-      "eval_runtime": 76.7476,
-      "eval_samples_per_second": 15.714,
-      "eval_steps_per_second": 1.967,
+      "epoch": 1.386518771331058,
+      "eval_loss": 0.9270105957984924,
+      "eval_runtime": 63.2619,
+      "eval_samples_per_second": 15.602,
+      "eval_steps_per_second": 1.96,
       "step": 13000
     },
     {
-      "epoch": 2.4866457911217537,
-      "grad_norm": 1.288072943687439,
-      "learning_rate": 1.891692761097808e-08,
-      "loss": 0.2015,
+      "epoch": 1.4398464163822526,
+      "grad_norm": 17.55596923828125,
+      "learning_rate": 3.200191979522184e-08,
+      "loss": 0.22,
       "step": 13500
     },
     {
-      "epoch": 2.4866457911217537,
-      "eval_loss": 0.9051916599273682,
-      "eval_runtime": 78.3909,
-      "eval_samples_per_second": 15.384,
-      "eval_steps_per_second": 1.926,
+      "epoch": 1.4398464163822526,
+      "eval_loss": 0.919725775718689,
+      "eval_runtime": 63.3459,
+      "eval_samples_per_second": 15.581,
+      "eval_steps_per_second": 1.958,
       "step": 13500
     },
     {
-      "epoch": 2.578743783385522,
-      "grad_norm": 5.835799311171286e-05,
-      "learning_rate": 1.7765702707680972e-08,
-      "loss": 0.1827,
+      "epoch": 1.493174061433447,
+      "grad_norm": 3.4780771732330322,
+      "learning_rate": 3.133532423208191e-08,
+      "loss": 0.1656,
       "step": 14000
     },
     {
-      "epoch": 2.578743783385522,
-      "eval_loss": 0.9026296734809875,
-      "eval_runtime": 76.6591,
-      "eval_samples_per_second": 15.732,
-      "eval_steps_per_second": 1.97,
+      "epoch": 1.493174061433447,
+      "eval_loss": 0.9148014783859253,
+      "eval_runtime": 62.257,
+      "eval_samples_per_second": 15.854,
+      "eval_steps_per_second": 1.992,
       "step": 14000
     },
     {
-      "epoch": 2.670841775649291,
-      "grad_norm": 30.217784881591797,
-      "learning_rate": 1.6614477804383865e-08,
-      "loss": 0.1435,
+      "epoch": 1.5465017064846416,
+      "grad_norm": 158.22390747070312,
+      "learning_rate": 3.066872866894198e-08,
+      "loss": 0.1968,
       "step": 14500
     },
     {
-      "epoch": 2.670841775649291,
-      "eval_loss": 0.899847686290741,
-      "eval_runtime": 76.742,
-      "eval_samples_per_second": 15.715,
-      "eval_steps_per_second": 1.968,
+      "epoch": 1.5465017064846416,
+      "eval_loss": 0.9096349477767944,
+      "eval_runtime": 63.3671,
+      "eval_samples_per_second": 15.576,
+      "eval_steps_per_second": 1.957,
       "step": 14500
     },
     {
-      "epoch": 2.7629397679130596,
-      "grad_norm": 202.30496215820312,
-      "learning_rate": 1.5463252901086757e-08,
-      "loss": 0.1541,
+      "epoch": 1.5998293515358362,
+      "grad_norm": 621.3922729492188,
+      "learning_rate": 3.0002133105802044e-08,
+      "loss": 0.1676,
       "step": 15000
     },
     {
-      "epoch": 2.7629397679130596,
-      "eval_loss": 0.896297812461853,
-      "eval_runtime": 78.5539,
-      "eval_samples_per_second": 15.353,
-      "eval_steps_per_second": 1.922,
+      "epoch": 1.5998293515358362,
+      "eval_loss": 0.9056846499443054,
+      "eval_runtime": 63.2369,
+      "eval_samples_per_second": 15.608,
+      "eval_steps_per_second": 1.961,
       "step": 15000
     },
     {
-      "epoch": 2.8550377601768284,
-      "grad_norm": 241.57069396972656,
-      "learning_rate": 1.4312027997789646e-08,
-      "loss": 0.1716,
+      "epoch": 1.6531569965870307,
+      "grad_norm": 26.23398208618164,
+      "learning_rate": 2.9335537542662113e-08,
+      "loss": 0.2074,
       "step": 15500
     },
     {
-      "epoch": 2.8550377601768284,
-      "eval_loss": 0.8935310244560242,
-      "eval_runtime": 76.9195,
-      "eval_samples_per_second": 15.679,
-      "eval_steps_per_second": 1.963,
+      "epoch": 1.6531569965870307,
+      "eval_loss": 0.8993550539016724,
+      "eval_runtime": 62.3767,
+      "eval_samples_per_second": 15.823,
+      "eval_steps_per_second": 1.988,
       "step": 15500
     },
     {
-      "epoch": 2.947135752440597,
-      "grad_norm": 0.25739413499832153,
-      "learning_rate": 1.316080309449254e-08,
-      "loss": 0.2056,
+      "epoch": 1.7064846416382253,
+      "grad_norm": 283.3314514160156,
+      "learning_rate": 2.8668941979522183e-08,
+      "loss": 0.1847,
       "step": 16000
     },
     {
-      "epoch": 2.947135752440597,
-      "eval_loss": 0.890542209148407,
-      "eval_runtime": 76.7268,
-      "eval_samples_per_second": 15.718,
-      "eval_steps_per_second": 1.968,
+      "epoch": 1.7064846416382253,
+      "eval_loss": 0.8954482078552246,
+      "eval_runtime": 63.2168,
+      "eval_samples_per_second": 15.613,
+      "eval_steps_per_second": 1.962,
       "step": 16000
     },
     {
-      "epoch": 3.0392337447043656,
-      "grad_norm": 86.5535659790039,
-      "learning_rate": 1.2009578191195431e-08,
-      "loss": 0.1843,
+      "epoch": 1.75981228668942,
+      "grad_norm": 12.151030540466309,
+      "learning_rate": 2.800234641638225e-08,
+      "loss": 0.1845,
       "step": 16500
     },
     {
-      "epoch": 3.0392337447043656,
-      "eval_loss": 0.8875145316123962,
-      "eval_runtime": 76.4556,
-      "eval_samples_per_second": 15.774,
-      "eval_steps_per_second": 1.975,
+      "epoch": 1.75981228668942,
+      "eval_loss": 0.8900338411331177,
+      "eval_runtime": 63.5581,
+      "eval_samples_per_second": 15.529,
+      "eval_steps_per_second": 1.951,
       "step": 16500
     },
     {
-      "epoch": 3.131331736968134,
-      "grad_norm": 0.05445564165711403,
-      "learning_rate": 1.0858353287898323e-08,
-      "loss": 0.1611,
+      "epoch": 1.8131399317406145,
+      "grad_norm": 0.7503092288970947,
+      "learning_rate": 2.7335750853242318e-08,
+      "loss": 0.1721,
       "step": 17000
     },
     {
-      "epoch": 3.131331736968134,
-      "eval_loss": 0.8858152627944946,
-      "eval_runtime": 77.4814,
-      "eval_samples_per_second": 15.565,
-      "eval_steps_per_second": 1.949,
+      "epoch": 1.8131399317406145,
+      "eval_loss": 0.8872589468955994,
+      "eval_runtime": 62.0282,
+      "eval_samples_per_second": 15.912,
+      "eval_steps_per_second": 1.999,
       "step": 17000
     },
     {
-      "epoch": 3.2240235814296243,
-      "grad_norm": 44.673702239990234,
-      "learning_rate": 9.699705232129698e-09,
-      "loss": 0.1568,
+      "epoch": 1.8664675767918089,
+      "grad_norm": 180.9355926513672,
+      "learning_rate": 2.6669155290102384e-08,
+      "loss": 0.2627,
       "step": 17500
     },
     {
-      "epoch": 3.2240235814296243,
-      "eval_loss": 0.7821062207221985,
-      "eval_runtime": 75.6502,
-      "eval_samples_per_second": 15.955,
-      "eval_steps_per_second": 1.996,
+      "epoch": 1.8664675767918089,
+      "eval_loss": 0.8810189962387085,
+      "eval_runtime": 62.7212,
+      "eval_samples_per_second": 15.736,
+      "eval_steps_per_second": 1.977,
       "step": 17500
     },
     {
-      "epoch": 3.316138540899042,
-      "grad_norm": 0.5617901086807251,
-      "learning_rate": 8.548268238761974e-09,
-      "loss": 0.1395,
+      "epoch": 1.9197952218430034,
+      "grad_norm": 0.10503419488668442,
+      "learning_rate": 2.600255972696246e-08,
+      "loss": 0.1623,
       "step": 18000
     },
     {
-      "epoch": 3.316138540899042,
-      "eval_loss": 0.7793659567832947,
-      "eval_runtime": 76.0283,
-      "eval_samples_per_second": 15.876,
-      "eval_steps_per_second": 1.986,
+      "epoch": 1.9197952218430034,
+      "eval_loss": 0.8773773908615112,
+      "eval_runtime": 62.9181,
+      "eval_samples_per_second": 15.687,
+      "eval_steps_per_second": 1.971,
       "step": 18000
     },
     {
-      "epoch": 3.4082535003684598,
-      "grad_norm": 6.4552321434021,
-      "learning_rate": 7.396831245394252e-09,
-      "loss": 0.1804,
+      "epoch": 1.9731228668941978,
+      "grad_norm": 477.76715087890625,
+      "learning_rate": 2.5335964163822526e-08,
+      "loss": 0.2162,
       "step": 18500
     },
     {
-      "epoch": 3.4082535003684598,
-      "eval_loss": 0.7778191566467285,
-      "eval_runtime": 76.0536,
-      "eval_samples_per_second": 15.87,
-      "eval_steps_per_second": 1.985,
+      "epoch": 1.9731228668941978,
+      "eval_loss": 0.8713328242301941,
+      "eval_runtime": 62.3477,
+      "eval_samples_per_second": 15.831,
+      "eval_steps_per_second": 1.989,
       "step": 18500
     },
     {
-      "epoch": 3.500368459837878,
-      "grad_norm": 0.31470850110054016,
-      "learning_rate": 6.2453942520265294e-09,
-      "loss": 0.1728,
+      "epoch": 2.0264505119453924,
+      "grad_norm": 222.06802368164062,
+      "learning_rate": 2.4669368600682595e-08,
+      "loss": 0.1802,
       "step": 19000
     },
     {
-      "epoch": 3.500368459837878,
-      "eval_loss": 0.7768829464912415,
-      "eval_runtime": 76.116,
-      "eval_samples_per_second": 15.857,
-      "eval_steps_per_second": 1.984,
+      "epoch": 2.0264505119453924,
+      "eval_loss": 0.8679063320159912,
+      "eval_runtime": 63.0768,
+      "eval_samples_per_second": 15.648,
+      "eval_steps_per_second": 1.966,
       "step": 19000
     },
     {
-      "epoch": 3.5924834193072956,
-      "grad_norm": 0.029128307476639748,
-      "learning_rate": 5.093957258658806e-09,
+      "epoch": 2.079778156996587,
+      "grad_norm": 48.63727569580078,
+      "learning_rate": 2.400277303754266e-08,
       "loss": 0.179,
       "step": 19500
     },
     {
-      "epoch": 3.5924834193072956,
-      "eval_loss": 0.7758385539054871,
-      "eval_runtime": 76.1889,
-      "eval_samples_per_second": 15.842,
-      "eval_steps_per_second": 1.982,
+      "epoch": 2.079778156996587,
+      "eval_loss": 0.86326664686203,
+      "eval_runtime": 63.3068,
+      "eval_samples_per_second": 15.591,
+      "eval_steps_per_second": 1.959,
       "step": 19500
     },
     {
-      "epoch": 3.6845983787767134,
-      "grad_norm": 0.0002703067730180919,
-      "learning_rate": 3.942520265291083e-09,
-      "loss": 0.179,
+      "epoch": 2.1331058020477816,
+      "grad_norm": 0.5827552080154419,
+      "learning_rate": 2.333617747440273e-08,
+      "loss": 0.1549,
       "step": 20000
     },
     {
-      "epoch": 3.6845983787767134,
-      "eval_loss": 0.7751660943031311,
-      "eval_runtime": 76.1566,
-      "eval_samples_per_second": 15.849,
-      "eval_steps_per_second": 1.983,
+      "epoch": 2.1331058020477816,
+      "eval_loss": 0.8606277108192444,
+      "eval_runtime": 62.3116,
+      "eval_samples_per_second": 15.84,
+      "eval_steps_per_second": 1.99,
       "step": 20000
     },
     {
-      "epoch": 3.776713338246131,
-      "grad_norm": 259.0263366699219,
-      "learning_rate": 2.79108327192336e-09,
-      "loss": 0.1454,
+      "epoch": 2.186433447098976,
+      "grad_norm": 0.060927845537662506,
+      "learning_rate": 2.2669581911262796e-08,
+      "loss": 0.1742,
       "step": 20500
     },
     {
-      "epoch": 3.776713338246131,
-      "eval_loss": 0.7746543884277344,
-      "eval_runtime": 76.3059,
-      "eval_samples_per_second": 15.818,
-      "eval_steps_per_second": 1.979,
+      "epoch": 2.186433447098976,
+      "eval_loss": 0.8584741353988647,
+      "eval_runtime": 63.1659,
+      "eval_samples_per_second": 15.626,
+      "eval_steps_per_second": 1.963,
       "step": 20500
     }
   ],
   "logging_steps": 500,
-  "max_steps": 21712,
+  "max_steps": 37504,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 4,
   "save_steps": 500,
@@ -641,7 +641,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 7375226080005630.0,
+  "total_flos": 7375765717532520.0,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null

checkpoint-20500/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:af19727b1180ae73a7b4839d8539085faed51bc7e6cd0858188fac5c6ef07636
+oid sha256:87e35dca972e13467c750106c19301b098fbd2918fda1087487e227bac3aca37
 size 4847