Training in progress, step 12500, checkpoint

checkpoint-12500/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fbf7ea7eaa0160158a2feae29daa0c2c8f60b9feb7d6d035737ac51596b8836c
+oid sha256:f6234a2adae32f48b2b8b9912a701ca49ececa3c45ff8c2b68b039f6c092b7ab
 size 1711848436

checkpoint-12500/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:adcc71b69741548226c7088ce276c13b82036991dd1639ceea0ec02776ab828e
+oid sha256:1ad3b1ec8045419093c133aa08e6c0e8f5916809a8b60fc9f05867f178e92df4
 size 3424043887

checkpoint-12500/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:44ce728a996d4cd2bb50f824c75c8ae07c806e6bf3c3d424b2413e32d621d8b9
+oid sha256:db030b30be6b93c99db6305a4a98097d55df00d054b7ea5674087844c194270b
 size 14503

checkpoint-12500/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e73baa6fbf1de8ba0ee3cdeef0186e8cee096a2f532addf4ee486a3edbe15fe0
+oid sha256:d506cc57af80b73b4c7db2e306c9ad0748fd18ecbbbf88cd414b1c91d19bee5e
 size 623

checkpoint-12500/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.5445749759674072,
-  "best_model_checkpoint": "./output/clip-finetuned-csu-p14-336-e3l57-l/checkpoint-12500",
-  "epoch": 2.3028739867354457,
+  "best_metric": 0.5412191152572632,
+  "best_model_checkpoint": "./output/clip-finetuned-csu-p14-336-e3l57-l/checkpoint-12000",
+  "epoch": 1.3331911262798635,
   "eval_steps": 500,
   "global_step": 12500,
   "is_hyper_param_search": false,
@@ -9,383 +9,383 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.09211495946941783,
-      "grad_norm": 2.6780805587768555,
-      "learning_rate": 4.84647506755097e-07,
-      "loss": 0.2705,
+      "epoch": 0.05332764505119454,
+      "grad_norm": 414.62841796875,
+      "learning_rate": 4.911120591581342e-07,
+      "loss": 0.3812,
       "step": 500
     },
     {
-      "epoch": 0.09211495946941783,
-      "eval_loss": 1.0680681467056274,
-      "eval_runtime": 74.5723,
-      "eval_samples_per_second": 16.186,
-      "eval_steps_per_second": 2.025,
+      "epoch": 0.05332764505119454,
+      "eval_loss": 1.1163015365600586,
+      "eval_runtime": 61.8962,
+      "eval_samples_per_second": 15.946,
+      "eval_steps_per_second": 2.003,
       "step": 500
     },
     {
-      "epoch": 0.18422991893883567,
-      "grad_norm": 8.301130294799805,
-      "learning_rate": 4.69295013510194e-07,
-      "loss": 0.2545,
+      "epoch": 0.10665529010238908,
+      "grad_norm": 34.08418655395508,
+      "learning_rate": 4.822241183162685e-07,
+      "loss": 0.2683,
       "step": 1000
     },
     {
-      "epoch": 0.18422991893883567,
-      "eval_loss": 0.9443553686141968,
-      "eval_runtime": 74.9643,
-      "eval_samples_per_second": 16.101,
-      "eval_steps_per_second": 2.014,
+      "epoch": 0.10665529010238908,
+      "eval_loss": 0.9684447646141052,
+      "eval_runtime": 61.821,
+      "eval_samples_per_second": 15.965,
+      "eval_steps_per_second": 2.006,
       "step": 1000
     },
     {
-      "epoch": 0.2763448784082535,
-      "grad_norm": 0.018898434937000275,
-      "learning_rate": 4.5394252026529107e-07,
-      "loss": 0.234,
+      "epoch": 0.1599829351535836,
+      "grad_norm": 350.4574890136719,
+      "learning_rate": 4.733361774744027e-07,
+      "loss": 0.2119,
       "step": 1500
     },
     {
-      "epoch": 0.2763448784082535,
-      "eval_loss": 0.876915693283081,
-      "eval_runtime": 75.6401,
-      "eval_samples_per_second": 15.957,
-      "eval_steps_per_second": 1.996,
+      "epoch": 0.1599829351535836,
+      "eval_loss": 0.9099885821342468,
+      "eval_runtime": 61.9606,
+      "eval_samples_per_second": 15.929,
+      "eval_steps_per_second": 2.001,
       "step": 1500
     },
     {
-      "epoch": 0.36845983787767134,
-      "grad_norm": 0.00010984255641233176,
-      "learning_rate": 4.385900270203881e-07,
-      "loss": 0.1539,
+      "epoch": 0.21331058020477817,
+      "grad_norm": 0.01787523180246353,
+      "learning_rate": 4.6444823663253695e-07,
+      "loss": 0.1889,
       "step": 2000
     },
     {
-      "epoch": 0.36845983787767134,
-      "eval_loss": 0.8415330648422241,
-      "eval_runtime": 76.9959,
-      "eval_samples_per_second": 15.676,
-      "eval_steps_per_second": 1.961,
+      "epoch": 0.21331058020477817,
+      "eval_loss": 0.8620074987411499,
+      "eval_runtime": 63.2824,
+      "eval_samples_per_second": 15.597,
+      "eval_steps_per_second": 1.959,
       "step": 2000
     },
     {
-      "epoch": 0.46057479734708917,
-      "grad_norm": 395.890625,
-      "learning_rate": 4.232375337754851e-07,
-      "loss": 0.1766,
+      "epoch": 0.2666382252559727,
+      "grad_norm": 118.66907501220703,
+      "learning_rate": 4.5556029579067116e-07,
+      "loss": 0.2071,
       "step": 2500
     },
     {
-      "epoch": 0.46057479734708917,
-      "eval_loss": 0.7660450339317322,
-      "eval_runtime": 76.733,
-      "eval_samples_per_second": 15.73,
-      "eval_steps_per_second": 1.968,
+      "epoch": 0.2666382252559727,
+      "eval_loss": 0.791786253452301,
+      "eval_runtime": 63.7932,
+      "eval_samples_per_second": 15.472,
+      "eval_steps_per_second": 1.944,
       "step": 2500
     },
     {
-      "epoch": 0.552689756816507,
-      "grad_norm": 363.6924743652344,
-      "learning_rate": 4.0788504053058217e-07,
-      "loss": 0.1679,
+      "epoch": 0.3199658703071672,
+      "grad_norm": 7.21087323185543e-09,
+      "learning_rate": 4.4667235494880547e-07,
+      "loss": 0.1588,
       "step": 3000
     },
     {
-      "epoch": 0.552689756816507,
-      "eval_loss": 0.7269155383110046,
-      "eval_runtime": 76.7928,
-      "eval_samples_per_second": 15.718,
-      "eval_steps_per_second": 1.966,
+      "epoch": 0.3199658703071672,
+      "eval_loss": 0.7657254338264465,
+      "eval_runtime": 62.4738,
+      "eval_samples_per_second": 15.799,
+      "eval_steps_per_second": 1.985,
       "step": 3000
     },
     {
-      "epoch": 0.6448047162859248,
-      "grad_norm": 28.914026260375977,
-      "learning_rate": 3.925325472856792e-07,
-      "loss": 0.1104,
+      "epoch": 0.37329351535836175,
+      "grad_norm": 0.0014836661284789443,
+      "learning_rate": 4.377844141069397e-07,
+      "loss": 0.1718,
       "step": 3500
     },
     {
-      "epoch": 0.6448047162859248,
-      "eval_loss": 0.7097663283348083,
-      "eval_runtime": 76.6897,
-      "eval_samples_per_second": 15.739,
-      "eval_steps_per_second": 1.969,
+      "epoch": 0.37329351535836175,
+      "eval_loss": 0.7610095739364624,
+      "eval_runtime": 63.9118,
+      "eval_samples_per_second": 15.443,
+      "eval_steps_per_second": 1.94,
       "step": 3500
     },
     {
-      "epoch": 0.7369196757553427,
-      "grad_norm": 116.85651397705078,
-      "learning_rate": 3.771800540407762e-07,
-      "loss": 0.1367,
+      "epoch": 0.42662116040955633,
+      "grad_norm": 3.2739710807800293,
+      "learning_rate": 4.2889647326507393e-07,
+      "loss": 0.1113,
       "step": 4000
     },
     {
-      "epoch": 0.7369196757553427,
-      "eval_loss": 0.6968969106674194,
-      "eval_runtime": 77.1872,
-      "eval_samples_per_second": 15.637,
-      "eval_steps_per_second": 1.956,
+      "epoch": 0.42662116040955633,
+      "eval_loss": 0.7458393573760986,
+      "eval_runtime": 62.1803,
+      "eval_samples_per_second": 15.873,
+      "eval_steps_per_second": 1.994,
       "step": 4000
     },
     {
-      "epoch": 0.8290346352247605,
-      "grad_norm": 391.0343322753906,
-      "learning_rate": 3.618275607958732e-07,
-      "loss": 0.1129,
+      "epoch": 0.47994880546075086,
+      "grad_norm": 1.788787841796875,
+      "learning_rate": 4.2000853242320814e-07,
+      "loss": 0.1313,
       "step": 4500
     },
     {
-      "epoch": 0.8290346352247605,
-      "eval_loss": 0.6776800751686096,
-      "eval_runtime": 75.0351,
-      "eval_samples_per_second": 16.086,
-      "eval_steps_per_second": 2.012,
+      "epoch": 0.47994880546075086,
+      "eval_loss": 0.7168156504631042,
+      "eval_runtime": 62.1477,
+      "eval_samples_per_second": 15.882,
+      "eval_steps_per_second": 1.995,
       "step": 4500
     },
     {
-      "epoch": 0.9211495946941783,
-      "grad_norm": 2.0514346033451147e-05,
-      "learning_rate": 3.464750675509703e-07,
-      "loss": 0.1125,
+      "epoch": 0.5332764505119454,
+      "grad_norm": 4.031916250823997e-05,
+      "learning_rate": 4.1112059158134245e-07,
+      "loss": 0.1649,
       "step": 5000
     },
     {
-      "epoch": 0.9211495946941783,
-      "eval_loss": 0.6658498048782349,
-      "eval_runtime": 77.111,
-      "eval_samples_per_second": 15.653,
-      "eval_steps_per_second": 1.958,
+      "epoch": 0.5332764505119454,
+      "eval_loss": 0.7019046545028687,
+      "eval_runtime": 62.2092,
+      "eval_samples_per_second": 15.866,
+      "eval_steps_per_second": 1.993,
       "step": 5000
     },
     {
-      "epoch": 1.013264554163596,
-      "grad_norm": 0.000568103976547718,
-      "learning_rate": 3.311225743060673e-07,
-      "loss": 0.1071,
+      "epoch": 0.58660409556314,
+      "grad_norm": 3.883213139488362e-06,
+      "learning_rate": 4.0223265073947665e-07,
+      "loss": 0.1245,
       "step": 5500
     },
     {
-      "epoch": 1.013264554163596,
-      "eval_loss": 0.6465296149253845,
-      "eval_runtime": 75.2951,
-      "eval_samples_per_second": 16.03,
-      "eval_steps_per_second": 2.005,
+      "epoch": 0.58660409556314,
+      "eval_loss": 0.6812178492546082,
+      "eval_runtime": 62.492,
+      "eval_samples_per_second": 15.794,
+      "eval_steps_per_second": 1.984,
       "step": 5500
     },
     {
-      "epoch": 1.105379513633014,
-      "grad_norm": 0.02698369137942791,
-      "learning_rate": 3.157700810611643e-07,
-      "loss": 0.0553,
+      "epoch": 0.6399317406143344,
+      "grad_norm": 80.17021942138672,
+      "learning_rate": 3.933447098976109e-07,
+      "loss": 0.1286,
       "step": 6000
     },
     {
-      "epoch": 1.105379513633014,
-      "eval_loss": 0.6357071995735168,
-      "eval_runtime": 76.6726,
-      "eval_samples_per_second": 15.742,
-      "eval_steps_per_second": 1.969,
+      "epoch": 0.6399317406143344,
+      "eval_loss": 0.6501584649085999,
+      "eval_runtime": 62.5236,
+      "eval_samples_per_second": 15.786,
+      "eval_steps_per_second": 1.983,
       "step": 6000
     },
     {
-      "epoch": 1.1974944731024317,
-      "grad_norm": 0.0005627564387395978,
-      "learning_rate": 3.0041758781626137e-07,
-      "loss": 0.0729,
+      "epoch": 0.693259385665529,
+      "grad_norm": 0.0001418297761119902,
+      "learning_rate": 3.8445676905574517e-07,
+      "loss": 0.1076,
       "step": 6500
     },
     {
-      "epoch": 1.1974944731024317,
-      "eval_loss": 0.628384530544281,
-      "eval_runtime": 75.6565,
-      "eval_samples_per_second": 15.954,
-      "eval_steps_per_second": 1.996,
+      "epoch": 0.693259385665529,
+      "eval_loss": 0.6153982281684875,
+      "eval_runtime": 62.2927,
+      "eval_samples_per_second": 15.845,
+      "eval_steps_per_second": 1.991,
       "step": 6500
     },
     {
-      "epoch": 1.2896094325718497,
-      "grad_norm": 4.7338733111246256e-07,
-      "learning_rate": 2.8506509457135833e-07,
-      "loss": 0.0476,
+      "epoch": 0.7465870307167235,
+      "grad_norm": 5.804526495012396e-07,
+      "learning_rate": 3.755688282138794e-07,
+      "loss": 0.1477,
       "step": 7000
     },
     {
-      "epoch": 1.2896094325718497,
-      "eval_loss": 0.6259528994560242,
-      "eval_runtime": 75.2674,
-      "eval_samples_per_second": 16.036,
-      "eval_steps_per_second": 2.006,
+      "epoch": 0.7465870307167235,
+      "eval_loss": 0.6118016839027405,
+      "eval_runtime": 63.1588,
+      "eval_samples_per_second": 15.627,
+      "eval_steps_per_second": 1.963,
       "step": 7000
     },
     {
-      "epoch": 1.3817243920412676,
-      "grad_norm": 28.43704605102539,
-      "learning_rate": 2.697126013264554e-07,
-      "loss": 0.0756,
+      "epoch": 0.7999146757679181,
+      "grad_norm": 2.66489315032959,
+      "learning_rate": 3.6668088737201363e-07,
+      "loss": 0.1315,
       "step": 7500
     },
     {
-      "epoch": 1.3817243920412676,
-      "eval_loss": 0.6102285981178284,
-      "eval_runtime": 76.1062,
-      "eval_samples_per_second": 15.859,
-      "eval_steps_per_second": 1.984,
+      "epoch": 0.7999146757679181,
+      "eval_loss": 0.601563036441803,
+      "eval_runtime": 62.9163,
+      "eval_samples_per_second": 15.688,
+      "eval_steps_per_second": 1.971,
       "step": 7500
     },
     {
-      "epoch": 1.4738393515106853,
-      "grad_norm": 0.0002958567056339234,
-      "learning_rate": 2.5436010808155247e-07,
-      "loss": 0.0797,
+      "epoch": 0.8532423208191127,
+      "grad_norm": 851.2547607421875,
+      "learning_rate": 3.5779294653014783e-07,
+      "loss": 0.1413,
       "step": 8000
     },
     {
-      "epoch": 1.4738393515106853,
-      "eval_loss": 0.6023213267326355,
-      "eval_runtime": 75.1644,
-      "eval_samples_per_second": 16.058,
-      "eval_steps_per_second": 2.009,
+      "epoch": 0.8532423208191127,
+      "eval_loss": 0.584888756275177,
+      "eval_runtime": 62.6735,
+      "eval_samples_per_second": 15.748,
+      "eval_steps_per_second": 1.979,
       "step": 8000
     },
     {
-      "epoch": 1.565954310980103,
-      "grad_norm": 0.0021068258211016655,
-      "learning_rate": 2.390076148366495e-07,
-      "loss": 0.0536,
+      "epoch": 0.9065699658703071,
+      "grad_norm": 6.041275628376752e-05,
+      "learning_rate": 3.4890500568828214e-07,
+      "loss": 0.124,
       "step": 8500
     },
     {
-      "epoch": 1.565954310980103,
-      "eval_loss": 0.5879408121109009,
-      "eval_runtime": 76.0196,
-      "eval_samples_per_second": 15.877,
-      "eval_steps_per_second": 1.986,
+      "epoch": 0.9065699658703071,
+      "eval_loss": 0.5766209363937378,
+      "eval_runtime": 63.3493,
+      "eval_samples_per_second": 15.58,
+      "eval_steps_per_second": 1.957,
       "step": 8500
     },
     {
-      "epoch": 1.658069270449521,
-      "grad_norm": 4.1152270569000393e-07,
-      "learning_rate": 2.236551215917465e-07,
-      "loss": 0.0784,
+      "epoch": 0.9598976109215017,
+      "grad_norm": 3.0901598930358887,
+      "learning_rate": 3.4001706484641635e-07,
+      "loss": 0.1215,
       "step": 9000
     },
     {
-      "epoch": 1.658069270449521,
-      "eval_loss": 0.5879626274108887,
-      "eval_runtime": 75.6453,
-      "eval_samples_per_second": 15.956,
-      "eval_steps_per_second": 1.996,
+      "epoch": 0.9598976109215017,
+      "eval_loss": 0.5558739304542542,
+      "eval_runtime": 63.0862,
+      "eval_samples_per_second": 15.645,
+      "eval_steps_per_second": 1.966,
       "step": 9000
     },
     {
-      "epoch": 1.750184229918939,
-      "grad_norm": 2.4213447913368213e-10,
-      "learning_rate": 2.083026283468435e-07,
-      "loss": 0.0703,
+      "epoch": 1.0132252559726962,
+      "grad_norm": 2509.93212890625,
+      "learning_rate": 3.311291240045506e-07,
+      "loss": 0.131,
       "step": 9500
     },
     {
-      "epoch": 1.750184229918939,
-      "eval_loss": 0.5665237307548523,
-      "eval_runtime": 74.9732,
-      "eval_samples_per_second": 16.099,
-      "eval_steps_per_second": 2.014,
+      "epoch": 1.0132252559726962,
+      "eval_loss": 0.563326358795166,
+      "eval_runtime": 62.3607,
+      "eval_samples_per_second": 15.827,
+      "eval_steps_per_second": 1.988,
       "step": 9500
     },
     {
-      "epoch": 1.8422991893883567,
-      "grad_norm": 5.074160799267702e-05,
-      "learning_rate": 1.9295013510194055e-07,
-      "loss": 0.0551,
+      "epoch": 1.0665529010238908,
+      "grad_norm": 32.05735778808594,
+      "learning_rate": 3.2224118316268486e-07,
+      "loss": 0.0348,
       "step": 10000
     },
     {
-      "epoch": 1.8422991893883567,
-      "eval_loss": 0.5670996308326721,
-      "eval_runtime": 75.902,
-      "eval_samples_per_second": 15.902,
-      "eval_steps_per_second": 1.989,
+      "epoch": 1.0665529010238908,
+      "eval_loss": 0.553093671798706,
+      "eval_runtime": 62.9895,
+      "eval_samples_per_second": 15.669,
+      "eval_steps_per_second": 1.969,
       "step": 10000
     },
     {
-      "epoch": 1.9344141488577744,
-      "grad_norm": 6.560769350016926e-08,
-      "learning_rate": 1.7759764185703757e-07,
-      "loss": 0.0852,
+      "epoch": 1.1198805460750854,
+      "grad_norm": 0.004324838053435087,
+      "learning_rate": 3.133532423208191e-07,
+      "loss": 0.0687,
       "step": 10500
     },
     {
-      "epoch": 1.9344141488577744,
-      "eval_loss": 0.5694836974143982,
-      "eval_runtime": 76.048,
-      "eval_samples_per_second": 15.872,
-      "eval_steps_per_second": 1.986,
+      "epoch": 1.1198805460750854,
+      "eval_loss": 0.5494681000709534,
+      "eval_runtime": 62.3211,
+      "eval_samples_per_second": 15.837,
+      "eval_steps_per_second": 1.99,
       "step": 10500
     },
     {
-      "epoch": 2.026529108327192,
-      "grad_norm": 4.498223916016286e-06,
-      "learning_rate": 1.6224514861213458e-07,
-      "loss": 0.0546,
+      "epoch": 1.17320819112628,
+      "grad_norm": 0.09391242265701294,
+      "learning_rate": 3.044653014789533e-07,
+      "loss": 0.0749,
       "step": 11000
     },
     {
-      "epoch": 2.026529108327192,
-      "eval_loss": 0.555758535861969,
-      "eval_runtime": 74.9986,
-      "eval_samples_per_second": 16.094,
-      "eval_steps_per_second": 2.013,
+      "epoch": 1.17320819112628,
+      "eval_loss": 0.5473881363868713,
+      "eval_runtime": 62.3454,
+      "eval_samples_per_second": 15.831,
+      "eval_steps_per_second": 1.989,
       "step": 11000
     },
     {
-      "epoch": 2.1186440677966103,
-      "grad_norm": 273.0765380859375,
-      "learning_rate": 1.4689265536723165e-07,
-      "loss": 0.0369,
+      "epoch": 1.2265358361774745,
+      "grad_norm": 1.9322777752850406e-11,
+      "learning_rate": 2.955773606370876e-07,
+      "loss": 0.1083,
       "step": 11500
     },
     {
-      "epoch": 2.1186440677966103,
-      "eval_loss": 0.5532959699630737,
-      "eval_runtime": 74.8225,
-      "eval_samples_per_second": 16.132,
-      "eval_steps_per_second": 2.018,
+      "epoch": 1.2265358361774745,
+      "eval_loss": 0.541618287563324,
+      "eval_runtime": 63.1868,
+      "eval_samples_per_second": 15.62,
+      "eval_steps_per_second": 1.962,
       "step": 11500
     },
     {
-      "epoch": 2.210759027266028,
-      "grad_norm": 0.9755889773368835,
-      "learning_rate": 1.3154016212232866e-07,
-      "loss": 0.0205,
+      "epoch": 1.2798634812286689,
+      "grad_norm": 0.0002804399700835347,
+      "learning_rate": 2.8668941979522184e-07,
+      "loss": 0.0485,
       "step": 12000
     },
     {
-      "epoch": 2.210759027266028,
-      "eval_loss": 0.5498348474502563,
-      "eval_runtime": 76.4506,
-      "eval_samples_per_second": 15.788,
-      "eval_steps_per_second": 1.975,
+      "epoch": 1.2798634812286689,
+      "eval_loss": 0.5412191152572632,
+      "eval_runtime": 62.9217,
+      "eval_samples_per_second": 15.686,
+      "eval_steps_per_second": 1.971,
       "step": 12000
     },
     {
-      "epoch": 2.3028739867354457,
-      "grad_norm": 2.7215068340301514,
-      "learning_rate": 1.1618766887742569e-07,
-      "loss": 0.0673,
+      "epoch": 1.3331911262798635,
+      "grad_norm": 8.90180089641035e-08,
+      "learning_rate": 2.778014789533561e-07,
+      "loss": 0.0449,
       "step": 12500
     },
     {
-      "epoch": 2.3028739867354457,
-      "eval_loss": 0.5445749759674072,
-      "eval_runtime": 74.9944,
-      "eval_samples_per_second": 16.095,
-      "eval_steps_per_second": 2.013,
+      "epoch": 1.3331911262798635,
+      "eval_loss": 0.5510581731796265,
+      "eval_runtime": 62.525,
+      "eval_samples_per_second": 15.786,
+      "eval_steps_per_second": 1.983,
       "step": 12500
     }
   ],
   "logging_steps": 500,
-  "max_steps": 16284,
+  "max_steps": 28128,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 3,
   "save_steps": 500,

checkpoint-12500/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ea34fd4333686a44a60d290faa4ae8e00df783131842d9ae3ef1b9e874a39349
+oid sha256:bd16a1abd112ae60a2700f590eb3bd73e96acd9a0a37b30bbfa3f8ee2dc23bc1
 size 4847