Training in progress, step 12500, checkpoint

checkpoint-12500/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9a2ead5c165a3a944bd3a1f3ad3c72ac22433b5b899c4bea935be8a55560ecdb
+oid sha256:fbf7ea7eaa0160158a2feae29daa0c2c8f60b9feb7d6d035737ac51596b8836c
 size 1711848436

checkpoint-12500/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3156b6f0adb5392782ab9441d215207598d4fe8bfa3033c3871faa4ed28364f9
+oid sha256:adcc71b69741548226c7088ce276c13b82036991dd1639ceea0ec02776ab828e
 size 3424043887

checkpoint-12500/trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.5504964590072632,
+  "best_metric": 0.5445749759674072,
   "best_model_checkpoint": "./output/clip-finetuned-csu-p14-336-e3l57-l/checkpoint-12500",
   "epoch": 2.3028739867354457,
   "eval_steps": 500,
@@ -10,377 +10,377 @@
   "log_history": [
     {
       "epoch": 0.09211495946941783,
-      "grad_norm": 3.083087682723999,
+      "grad_norm": 2.6780805587768555,
       "learning_rate": 4.84647506755097e-07,
-      "loss": 0.2708,
+      "loss": 0.2705,
       "step": 500
     },
     {
       "epoch": 0.09211495946941783,
-      "eval_loss": 1.0631108283996582,
-      "eval_runtime": 74.1964,
-      "eval_samples_per_second": 16.268,
-      "eval_steps_per_second": 2.035,
+      "eval_loss": 1.0680681467056274,
+      "eval_runtime": 74.5723,
+      "eval_samples_per_second": 16.186,
+      "eval_steps_per_second": 2.025,
       "step": 500
     },
     {
       "epoch": 0.18422991893883567,
-      "grad_norm": 7.726487159729004,
+      "grad_norm": 8.301130294799805,
       "learning_rate": 4.69295013510194e-07,
-      "loss": 0.251,
+      "loss": 0.2545,
       "step": 1000
     },
     {
       "epoch": 0.18422991893883567,
-      "eval_loss": 0.942773699760437,
-      "eval_runtime": 74.6047,
-      "eval_samples_per_second": 16.179,
-      "eval_steps_per_second": 2.024,
+      "eval_loss": 0.9443553686141968,
+      "eval_runtime": 74.9643,
+      "eval_samples_per_second": 16.101,
+      "eval_steps_per_second": 2.014,
       "step": 1000
     },
     {
       "epoch": 0.2763448784082535,
-      "grad_norm": 0.02718021161854267,
+      "grad_norm": 0.018898434937000275,
       "learning_rate": 4.5394252026529107e-07,
-      "loss": 0.235,
+      "loss": 0.234,
       "step": 1500
     },
     {
       "epoch": 0.2763448784082535,
-      "eval_loss": 0.8716810345649719,
-      "eval_runtime": 75.03,
-      "eval_samples_per_second": 16.087,
-      "eval_steps_per_second": 2.013,
+      "eval_loss": 0.876915693283081,
+      "eval_runtime": 75.6401,
+      "eval_samples_per_second": 15.957,
+      "eval_steps_per_second": 1.996,
       "step": 1500
     },
     {
       "epoch": 0.36845983787767134,
-      "grad_norm": 0.0001215290030813776,
+      "grad_norm": 0.00010984255641233176,
       "learning_rate": 4.385900270203881e-07,
-      "loss": 0.1529,
+      "loss": 0.1539,
       "step": 2000
     },
     {
       "epoch": 0.36845983787767134,
-      "eval_loss": 0.8318113088607788,
-      "eval_runtime": 76.6917,
-      "eval_samples_per_second": 15.738,
-      "eval_steps_per_second": 1.969,
+      "eval_loss": 0.8415330648422241,
+      "eval_runtime": 76.9959,
+      "eval_samples_per_second": 15.676,
+      "eval_steps_per_second": 1.961,
       "step": 2000
     },
     {
       "epoch": 0.46057479734708917,
-      "grad_norm": 419.545166015625,
+      "grad_norm": 395.890625,
       "learning_rate": 4.232375337754851e-07,
-      "loss": 0.1781,
+      "loss": 0.1766,
       "step": 2500
     },
     {
       "epoch": 0.46057479734708917,
-      "eval_loss": 0.7549223303794861,
-      "eval_runtime": 75.2036,
-      "eval_samples_per_second": 16.05,
-      "eval_steps_per_second": 2.008,
+      "eval_loss": 0.7660450339317322,
+      "eval_runtime": 76.733,
+      "eval_samples_per_second": 15.73,
+      "eval_steps_per_second": 1.968,
       "step": 2500
     },
     {
       "epoch": 0.552689756816507,
-      "grad_norm": 368.7484436035156,
+      "grad_norm": 363.6924743652344,
       "learning_rate": 4.0788504053058217e-07,
-      "loss": 0.1681,
+      "loss": 0.1679,
       "step": 3000
     },
     {
       "epoch": 0.552689756816507,
-      "eval_loss": 0.7217888236045837,
-      "eval_runtime": 76.0606,
-      "eval_samples_per_second": 15.869,
-      "eval_steps_per_second": 1.985,
+      "eval_loss": 0.7269155383110046,
+      "eval_runtime": 76.7928,
+      "eval_samples_per_second": 15.718,
+      "eval_steps_per_second": 1.966,
       "step": 3000
     },
     {
       "epoch": 0.6448047162859248,
-      "grad_norm": 20.03949737548828,
+      "grad_norm": 28.914026260375977,
       "learning_rate": 3.925325472856792e-07,
-      "loss": 0.1064,
+      "loss": 0.1104,
       "step": 3500
     },
     {
       "epoch": 0.6448047162859248,
-      "eval_loss": 0.7048410177230835,
-      "eval_runtime": 76.0578,
-      "eval_samples_per_second": 15.87,
-      "eval_steps_per_second": 1.985,
+      "eval_loss": 0.7097663283348083,
+      "eval_runtime": 76.6897,
+      "eval_samples_per_second": 15.739,
+      "eval_steps_per_second": 1.969,
       "step": 3500
     },
     {
       "epoch": 0.7369196757553427,
-      "grad_norm": 363.6124267578125,
+      "grad_norm": 116.85651397705078,
       "learning_rate": 3.771800540407762e-07,
-      "loss": 0.1357,
+      "loss": 0.1367,
       "step": 4000
     },
     {
       "epoch": 0.7369196757553427,
-      "eval_loss": 0.6961681246757507,
-      "eval_runtime": 76.3574,
-      "eval_samples_per_second": 15.807,
-      "eval_steps_per_second": 1.978,
+      "eval_loss": 0.6968969106674194,
+      "eval_runtime": 77.1872,
+      "eval_samples_per_second": 15.637,
+      "eval_steps_per_second": 1.956,
       "step": 4000
     },
     {
       "epoch": 0.8290346352247605,
-      "grad_norm": 494.03570556640625,
+      "grad_norm": 391.0343322753906,
       "learning_rate": 3.618275607958732e-07,
-      "loss": 0.1098,
+      "loss": 0.1129,
       "step": 4500
     },
     {
       "epoch": 0.8290346352247605,
-      "eval_loss": 0.6778352856636047,
-      "eval_runtime": 76.4844,
-      "eval_samples_per_second": 15.781,
-      "eval_steps_per_second": 1.974,
+      "eval_loss": 0.6776800751686096,
+      "eval_runtime": 75.0351,
+      "eval_samples_per_second": 16.086,
+      "eval_steps_per_second": 2.012,
       "step": 4500
     },
     {
       "epoch": 0.9211495946941783,
-      "grad_norm": 4.2969353671651334e-05,
+      "grad_norm": 2.0514346033451147e-05,
       "learning_rate": 3.464750675509703e-07,
-      "loss": 0.1142,
+      "loss": 0.1125,
       "step": 5000
     },
     {
       "epoch": 0.9211495946941783,
-      "eval_loss": 0.6656659245491028,
-      "eval_runtime": 76.6475,
-      "eval_samples_per_second": 15.747,
-      "eval_steps_per_second": 1.97,
+      "eval_loss": 0.6658498048782349,
+      "eval_runtime": 77.111,
+      "eval_samples_per_second": 15.653,
+      "eval_steps_per_second": 1.958,
       "step": 5000
     },
     {
       "epoch": 1.013264554163596,
-      "grad_norm": 0.0010161151876673102,
+      "grad_norm": 0.000568103976547718,
       "learning_rate": 3.311225743060673e-07,
-      "loss": 0.1113,
+      "loss": 0.1071,
       "step": 5500
     },
     {
       "epoch": 1.013264554163596,
-      "eval_loss": 0.6430702209472656,
-      "eval_runtime": 76.4915,
-      "eval_samples_per_second": 15.78,
-      "eval_steps_per_second": 1.974,
+      "eval_loss": 0.6465296149253845,
+      "eval_runtime": 75.2951,
+      "eval_samples_per_second": 16.03,
+      "eval_steps_per_second": 2.005,
       "step": 5500
     },
     {
       "epoch": 1.105379513633014,
-      "grad_norm": 0.05515381693840027,
+      "grad_norm": 0.02698369137942791,
       "learning_rate": 3.157700810611643e-07,
-      "loss": 0.0572,
+      "loss": 0.0553,
       "step": 6000
     },
     {
       "epoch": 1.105379513633014,
-      "eval_loss": 0.636746883392334,
-      "eval_runtime": 76.0169,
-      "eval_samples_per_second": 15.878,
-      "eval_steps_per_second": 1.986,
+      "eval_loss": 0.6357071995735168,
+      "eval_runtime": 76.6726,
+      "eval_samples_per_second": 15.742,
+      "eval_steps_per_second": 1.969,
       "step": 6000
     },
     {
       "epoch": 1.1974944731024317,
-      "grad_norm": 0.0038496837951242924,
+      "grad_norm": 0.0005627564387395978,
       "learning_rate": 3.0041758781626137e-07,
-      "loss": 0.0746,
+      "loss": 0.0729,
       "step": 6500
     },
     {
       "epoch": 1.1974944731024317,
-      "eval_loss": 0.6261336207389832,
-      "eval_runtime": 76.1916,
-      "eval_samples_per_second": 15.842,
-      "eval_steps_per_second": 1.982,
+      "eval_loss": 0.628384530544281,
+      "eval_runtime": 75.6565,
+      "eval_samples_per_second": 15.954,
+      "eval_steps_per_second": 1.996,
       "step": 6500
     },
     {
       "epoch": 1.2896094325718497,
-      "grad_norm": 2.594869386030041e-07,
+      "grad_norm": 4.7338733111246256e-07,
       "learning_rate": 2.8506509457135833e-07,
-      "loss": 0.0494,
+      "loss": 0.0476,
       "step": 7000
     },
     {
       "epoch": 1.2896094325718497,
-      "eval_loss": 0.6245251893997192,
-      "eval_runtime": 75.9712,
-      "eval_samples_per_second": 15.888,
-      "eval_steps_per_second": 1.988,
+      "eval_loss": 0.6259528994560242,
+      "eval_runtime": 75.2674,
+      "eval_samples_per_second": 16.036,
+      "eval_steps_per_second": 2.006,
       "step": 7000
     },
     {
       "epoch": 1.3817243920412676,
-      "grad_norm": 25.23366928100586,
+      "grad_norm": 28.43704605102539,
       "learning_rate": 2.697126013264554e-07,
-      "loss": 0.0788,
+      "loss": 0.0756,
       "step": 7500
     },
     {
       "epoch": 1.3817243920412676,
-      "eval_loss": 0.611993670463562,
-      "eval_runtime": 75.1885,
-      "eval_samples_per_second": 16.053,
-      "eval_steps_per_second": 2.008,
+      "eval_loss": 0.6102285981178284,
+      "eval_runtime": 76.1062,
+      "eval_samples_per_second": 15.859,
+      "eval_steps_per_second": 1.984,
       "step": 7500
     },
     {
       "epoch": 1.4738393515106853,
-      "grad_norm": 0.0002172550739487633,
+      "grad_norm": 0.0002958567056339234,
       "learning_rate": 2.5436010808155247e-07,
-      "loss": 0.0808,
+      "loss": 0.0797,
       "step": 8000
     },
     {
       "epoch": 1.4738393515106853,
-      "eval_loss": 0.6011174917221069,
-      "eval_runtime": 76.2572,
-      "eval_samples_per_second": 15.828,
-      "eval_steps_per_second": 1.98,
+      "eval_loss": 0.6023213267326355,
+      "eval_runtime": 75.1644,
+      "eval_samples_per_second": 16.058,
+      "eval_steps_per_second": 2.009,
       "step": 8000
     },
     {
       "epoch": 1.565954310980103,
-      "grad_norm": 0.032347481697797775,
+      "grad_norm": 0.0021068258211016655,
       "learning_rate": 2.390076148366495e-07,
       "loss": 0.0536,
       "step": 8500
     },
     {
       "epoch": 1.565954310980103,
-      "eval_loss": 0.5893377661705017,
-      "eval_runtime": 75.429,
-      "eval_samples_per_second": 16.002,
-      "eval_steps_per_second": 2.002,
+      "eval_loss": 0.5879408121109009,
+      "eval_runtime": 76.0196,
+      "eval_samples_per_second": 15.877,
+      "eval_steps_per_second": 1.986,
       "step": 8500
     },
     {
       "epoch": 1.658069270449521,
-      "grad_norm": 2.7040863415095373e-07,
+      "grad_norm": 4.1152270569000393e-07,
       "learning_rate": 2.236551215917465e-07,
-      "loss": 0.0869,
+      "loss": 0.0784,
       "step": 9000
     },
     {
       "epoch": 1.658069270449521,
-      "eval_loss": 0.59163498878479,
-      "eval_runtime": 75.4027,
-      "eval_samples_per_second": 16.007,
-      "eval_steps_per_second": 2.003,
+      "eval_loss": 0.5879626274108887,
+      "eval_runtime": 75.6453,
+      "eval_samples_per_second": 15.956,
+      "eval_steps_per_second": 1.996,
       "step": 9000
     },
     {
       "epoch": 1.750184229918939,
-      "grad_norm": 3.4501523166197146e-11,
+      "grad_norm": 2.4213447913368213e-10,
       "learning_rate": 2.083026283468435e-07,
-      "loss": 0.0752,
+      "loss": 0.0703,
       "step": 9500
     },
     {
       "epoch": 1.750184229918939,
-      "eval_loss": 0.5707471370697021,
-      "eval_runtime": 76.0154,
-      "eval_samples_per_second": 15.878,
-      "eval_steps_per_second": 1.986,
+      "eval_loss": 0.5665237307548523,
+      "eval_runtime": 74.9732,
+      "eval_samples_per_second": 16.099,
+      "eval_steps_per_second": 2.014,
       "step": 9500
     },
     {
       "epoch": 1.8422991893883567,
-      "grad_norm": 0.00039183301851153374,
+      "grad_norm": 5.074160799267702e-05,
       "learning_rate": 1.9295013510194055e-07,
-      "loss": 0.0577,
+      "loss": 0.0551,
       "step": 10000
     },
     {
       "epoch": 1.8422991893883567,
-      "eval_loss": 0.5678250789642334,
-      "eval_runtime": 75.8084,
-      "eval_samples_per_second": 15.922,
-      "eval_steps_per_second": 1.992,
+      "eval_loss": 0.5670996308326721,
+      "eval_runtime": 75.902,
+      "eval_samples_per_second": 15.902,
+      "eval_steps_per_second": 1.989,
       "step": 10000
     },
     {
       "epoch": 1.9344141488577744,
-      "grad_norm": 1.0438248487787405e-08,
+      "grad_norm": 6.560769350016926e-08,
       "learning_rate": 1.7759764185703757e-07,
-      "loss": 0.0891,
+      "loss": 0.0852,
       "step": 10500
     },
     {
       "epoch": 1.9344141488577744,
-      "eval_loss": 0.563089907169342,
-      "eval_runtime": 75.3868,
-      "eval_samples_per_second": 16.011,
-      "eval_steps_per_second": 2.003,
+      "eval_loss": 0.5694836974143982,
+      "eval_runtime": 76.048,
+      "eval_samples_per_second": 15.872,
+      "eval_steps_per_second": 1.986,
       "step": 10500
     },
     {
       "epoch": 2.026529108327192,
-      "grad_norm": 1.162303306045942e-06,
+      "grad_norm": 4.498223916016286e-06,
       "learning_rate": 1.6224514861213458e-07,
-      "loss": 0.0559,
+      "loss": 0.0546,
       "step": 11000
     },
     {
       "epoch": 2.026529108327192,
-      "eval_loss": 0.5547806620597839,
-      "eval_runtime": 76.4427,
-      "eval_samples_per_second": 15.79,
-      "eval_steps_per_second": 1.975,
+      "eval_loss": 0.555758535861969,
+      "eval_runtime": 74.9986,
+      "eval_samples_per_second": 16.094,
+      "eval_steps_per_second": 2.013,
       "step": 11000
     },
     {
       "epoch": 2.1186440677966103,
-      "grad_norm": 123.83905029296875,
+      "grad_norm": 273.0765380859375,
       "learning_rate": 1.4689265536723165e-07,
-      "loss": 0.0385,
+      "loss": 0.0369,
       "step": 11500
     },
     {
       "epoch": 2.1186440677966103,
-      "eval_loss": 0.5536245703697205,
-      "eval_runtime": 75.6886,
-      "eval_samples_per_second": 15.947,
-      "eval_steps_per_second": 1.995,
+      "eval_loss": 0.5532959699630737,
+      "eval_runtime": 74.8225,
+      "eval_samples_per_second": 16.132,
+      "eval_steps_per_second": 2.018,
       "step": 11500
     },
     {
       "epoch": 2.210759027266028,
-      "grad_norm": 1.234429121017456,
+      "grad_norm": 0.9755889773368835,
       "learning_rate": 1.3154016212232866e-07,
-      "loss": 0.0185,
+      "loss": 0.0205,
       "step": 12000
     },
     {
       "epoch": 2.210759027266028,
-      "eval_loss": 0.5518601536750793,
-      "eval_runtime": 75.2276,
-      "eval_samples_per_second": 16.045,
-      "eval_steps_per_second": 2.007,
+      "eval_loss": 0.5498348474502563,
+      "eval_runtime": 76.4506,
+      "eval_samples_per_second": 15.788,
+      "eval_steps_per_second": 1.975,
       "step": 12000
     },
     {
       "epoch": 2.3028739867354457,
-      "grad_norm": 3.730309009552002,
+      "grad_norm": 2.7215068340301514,
       "learning_rate": 1.1618766887742569e-07,
-      "loss": 0.0642,
+      "loss": 0.0673,
       "step": 12500
     },
     {
       "epoch": 2.3028739867354457,
-      "eval_loss": 0.5504964590072632,
-      "eval_runtime": 76.2015,
-      "eval_samples_per_second": 15.84,
-      "eval_steps_per_second": 1.982,
+      "eval_loss": 0.5445749759674072,
+      "eval_runtime": 74.9944,
+      "eval_samples_per_second": 16.095,
+      "eval_steps_per_second": 2.013,
       "step": 12500
     }
   ],

checkpoint-12500/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5c9589d95f7287a546fe28eed1162f0176f1b95b18274e91ad5fc36f7db12aeb
+oid sha256:ea34fd4333686a44a60d290faa4ae8e00df783131842d9ae3ef1b9e874a39349
 size 4847