Training in progress, step 15500, checkpoint

checkpoint-15500/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:46af8115c5637c2833a446b6f4f0528423865bbaed3308e2c2c1da49dcf747b8
+oid sha256:94fc9db36eef13d3e78982c75dc95410e482dde4724d876650b57fccc5b886ea
 size 1711848436

checkpoint-15500/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9e4ab866d5211edd6bd88f654c366d7e10fe517d874b3dd435fb824cd4a47f8f
+oid sha256:54ce55a49cce53855d4ba1adb5c3d01ab73528a8fe165e16e76e1910b67f007b
 size 3424043887

checkpoint-15500/trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.8834386467933655,
+  "best_metric": 0.8826532959938049,
   "best_model_checkpoint": "./output/clip-finetuned-csu-p14-336-e4l58-l/checkpoint-15500",
   "epoch": 2.855563743551953,
   "eval_steps": 500,
@@ -10,467 +10,467 @@
   "log_history": [
     {
       "epoch": 0.09211495946941783,
-      "grad_norm": 20.28152847290039,
+      "grad_norm": 20.04043960571289,
       "learning_rate": 4.8848563006632274e-08,
       "loss": 0.3758,
       "step": 500
     },
     {
       "epoch": 0.09211495946941783,
-      "eval_loss": 1.418367862701416,
-      "eval_runtime": 73.4439,
-      "eval_samples_per_second": 16.434,
-      "eval_steps_per_second": 2.056,
+      "eval_loss": 1.4184927940368652,
+      "eval_runtime": 74.3562,
+      "eval_samples_per_second": 16.233,
+      "eval_steps_per_second": 2.031,
       "step": 500
     },
     {
       "epoch": 0.18422991893883567,
-      "grad_norm": 47.29538345336914,
+      "grad_norm": 74.25025939941406,
       "learning_rate": 4.7697126013264556e-08,
-      "loss": 0.4104,
+      "loss": 0.4103,
       "step": 1000
     },
     {
       "epoch": 0.18422991893883567,
-      "eval_loss": 1.3491536378860474,
-      "eval_runtime": 74.3404,
-      "eval_samples_per_second": 16.236,
-      "eval_steps_per_second": 2.031,
+      "eval_loss": 1.3500770330429077,
+      "eval_runtime": 75.7013,
+      "eval_samples_per_second": 15.944,
+      "eval_steps_per_second": 1.995,
       "step": 1000
     },
     {
       "epoch": 0.2763448784082535,
-      "grad_norm": 0.5450155138969421,
+      "grad_norm": 0.5102410912513733,
       "learning_rate": 4.6545689019896826e-08,
-      "loss": 0.4325,
+      "loss": 0.433,
       "step": 1500
     },
     {
       "epoch": 0.2763448784082535,
-      "eval_loss": 1.2877168655395508,
-      "eval_runtime": 75.3303,
-      "eval_samples_per_second": 16.023,
-      "eval_steps_per_second": 2.005,
+      "eval_loss": 1.2885024547576904,
+      "eval_runtime": 75.197,
+      "eval_samples_per_second": 16.051,
+      "eval_steps_per_second": 2.008,
       "step": 1500
     },
     {
       "epoch": 0.36845983787767134,
-      "grad_norm": 0.11449883133172989,
+      "grad_norm": 0.1667678952217102,
       "learning_rate": 4.539425202652911e-08,
-      "loss": 0.3423,
+      "loss": 0.3424,
       "step": 2000
     },
     {
       "epoch": 0.36845983787767134,
-      "eval_loss": 1.2384389638900757,
-      "eval_runtime": 76.113,
-      "eval_samples_per_second": 15.858,
-      "eval_steps_per_second": 1.984,
+      "eval_loss": 1.239119052886963,
+      "eval_runtime": 75.387,
+      "eval_samples_per_second": 16.011,
+      "eval_steps_per_second": 2.003,
       "step": 2000
     },
     {
       "epoch": 0.46057479734708917,
-      "grad_norm": 389.7380065917969,
+      "grad_norm": 343.8175048828125,
       "learning_rate": 4.4242815033161385e-08,
       "loss": 0.3645,
       "step": 2500
     },
     {
       "epoch": 0.46057479734708917,
-      "eval_loss": 1.1896915435791016,
-      "eval_runtime": 76.23,
-      "eval_samples_per_second": 15.834,
-      "eval_steps_per_second": 1.981,
+      "eval_loss": 1.1902339458465576,
+      "eval_runtime": 75.9899,
+      "eval_samples_per_second": 15.884,
+      "eval_steps_per_second": 1.987,
       "step": 2500
     },
     {
       "epoch": 0.552689756816507,
-      "grad_norm": 340.1103820800781,
+      "grad_norm": 305.5626525878906,
       "learning_rate": 4.309137803979366e-08,
-      "loss": 0.3171,
+      "loss": 0.3172,
       "step": 3000
     },
     {
       "epoch": 0.552689756816507,
-      "eval_loss": 1.149291753768921,
-      "eval_runtime": 76.1662,
-      "eval_samples_per_second": 15.847,
-      "eval_steps_per_second": 1.983,
+      "eval_loss": 1.1506118774414062,
+      "eval_runtime": 74.8655,
+      "eval_samples_per_second": 16.122,
+      "eval_steps_per_second": 2.017,
       "step": 3000
     },
     {
       "epoch": 0.6448047162859248,
-      "grad_norm": 196.7362518310547,
+      "grad_norm": 65.59048461914062,
       "learning_rate": 4.193994104642594e-08,
-      "loss": 0.275,
+      "loss": 0.2751,
       "step": 3500
     },
     {
       "epoch": 0.6448047162859248,
-      "eval_loss": 1.1162768602371216,
-      "eval_runtime": 74.6569,
-      "eval_samples_per_second": 16.167,
-      "eval_steps_per_second": 2.023,
+      "eval_loss": 1.1169357299804688,
+      "eval_runtime": 76.2936,
+      "eval_samples_per_second": 15.82,
+      "eval_steps_per_second": 1.979,
       "step": 3500
     },
     {
       "epoch": 0.7369196757553427,
-      "grad_norm": 16.604211807250977,
+      "grad_norm": 18.106201171875,
       "learning_rate": 4.078850405305821e-08,
-      "loss": 0.2907,
+      "loss": 0.2919,
       "step": 4000
     },
     {
       "epoch": 0.7369196757553427,
-      "eval_loss": 1.091786503791809,
-      "eval_runtime": 76.3333,
-      "eval_samples_per_second": 15.812,
-      "eval_steps_per_second": 1.978,
+      "eval_loss": 1.0920671224594116,
+      "eval_runtime": 74.9101,
+      "eval_samples_per_second": 16.113,
+      "eval_steps_per_second": 2.016,
       "step": 4000
     },
     {
       "epoch": 0.8290346352247605,
-      "grad_norm": 527.0987548828125,
+      "grad_norm": 506.40570068359375,
       "learning_rate": 3.9637067059690496e-08,
-      "loss": 0.2581,
+      "loss": 0.2583,
       "step": 4500
     },
     {
       "epoch": 0.8290346352247605,
-      "eval_loss": 1.0729899406433105,
-      "eval_runtime": 74.4912,
-      "eval_samples_per_second": 16.203,
-      "eval_steps_per_second": 2.027,
+      "eval_loss": 1.0721209049224854,
+      "eval_runtime": 75.8053,
+      "eval_samples_per_second": 15.922,
+      "eval_steps_per_second": 1.992,
       "step": 4500
     },
     {
       "epoch": 0.9211495946941783,
-      "grad_norm": 1.148181676864624,
+      "grad_norm": 1.1508910655975342,
       "learning_rate": 3.848563006632277e-08,
-      "loss": 0.268,
+      "loss": 0.2679,
       "step": 5000
     },
     {
       "epoch": 0.9211495946941783,
-      "eval_loss": 1.052911400794983,
-      "eval_runtime": 74.9586,
-      "eval_samples_per_second": 16.102,
-      "eval_steps_per_second": 2.014,
+      "eval_loss": 1.0519349575042725,
+      "eval_runtime": 75.226,
+      "eval_samples_per_second": 16.045,
+      "eval_steps_per_second": 2.007,
       "step": 5000
     },
     {
       "epoch": 1.013264554163596,
-      "grad_norm": 0.1680648922920227,
+      "grad_norm": 0.1393290013074875,
       "learning_rate": 3.733419307295505e-08,
-      "loss": 0.2467,
+      "loss": 0.2472,
       "step": 5500
     },
     {
       "epoch": 1.013264554163596,
-      "eval_loss": 1.036412239074707,
-      "eval_runtime": 74.707,
-      "eval_samples_per_second": 16.156,
-      "eval_steps_per_second": 2.021,
+      "eval_loss": 1.0355629920959473,
+      "eval_runtime": 74.6535,
+      "eval_samples_per_second": 16.168,
+      "eval_steps_per_second": 2.023,
       "step": 5500
     },
     {
       "epoch": 1.105379513633014,
-      "grad_norm": 269.06585693359375,
+      "grad_norm": 200.34664916992188,
       "learning_rate": 3.6182756079587324e-08,
-      "loss": 0.2594,
+      "loss": 0.26,
       "step": 6000
     },
     {
       "epoch": 1.105379513633014,
-      "eval_loss": 1.0190415382385254,
-      "eval_runtime": 74.5657,
-      "eval_samples_per_second": 16.187,
-      "eval_steps_per_second": 2.025,
+      "eval_loss": 1.0176944732666016,
+      "eval_runtime": 74.4374,
+      "eval_samples_per_second": 16.215,
+      "eval_steps_per_second": 2.029,
       "step": 6000
     },
     {
       "epoch": 1.1974944731024317,
-      "grad_norm": 457.15850830078125,
+      "grad_norm": 431.1804504394531,
       "learning_rate": 3.50313190862196e-08,
       "loss": 0.2153,
       "step": 6500
     },
     {
       "epoch": 1.1974944731024317,
-      "eval_loss": 1.0054280757904053,
-      "eval_runtime": 74.5269,
-      "eval_samples_per_second": 16.195,
-      "eval_steps_per_second": 2.026,
+      "eval_loss": 1.0044586658477783,
+      "eval_runtime": 74.3247,
+      "eval_samples_per_second": 16.24,
+      "eval_steps_per_second": 2.032,
       "step": 6500
     },
     {
       "epoch": 1.2896094325718497,
-      "grad_norm": 0.5831326842308044,
+      "grad_norm": 0.6204941272735596,
       "learning_rate": 3.387988209285188e-08,
-      "loss": 0.1792,
+      "loss": 0.1791,
       "step": 7000
     },
     {
       "epoch": 1.2896094325718497,
-      "eval_loss": 0.9938780665397644,
-      "eval_runtime": 74.5283,
-      "eval_samples_per_second": 16.195,
-      "eval_steps_per_second": 2.026,
+      "eval_loss": 0.9926707148551941,
+      "eval_runtime": 75.7317,
+      "eval_samples_per_second": 15.938,
+      "eval_steps_per_second": 1.994,
       "step": 7000
     },
     {
       "epoch": 1.3817243920412676,
-      "grad_norm": 54.12948226928711,
+      "grad_norm": 52.27273178100586,
       "learning_rate": 3.272844509948415e-08,
-      "loss": 0.208,
+      "loss": 0.2082,
       "step": 7500
     },
     {
       "epoch": 1.3817243920412676,
-      "eval_loss": 0.9816781282424927,
-      "eval_runtime": 76.244,
-      "eval_samples_per_second": 15.831,
-      "eval_steps_per_second": 1.98,
+      "eval_loss": 0.9803809523582458,
+      "eval_runtime": 74.9936,
+      "eval_samples_per_second": 16.095,
+      "eval_steps_per_second": 2.014,
       "step": 7500
     },
     {
       "epoch": 1.4738393515106853,
-      "grad_norm": 0.09778845310211182,
+      "grad_norm": 0.0976635068655014,
       "learning_rate": 3.1577008106116435e-08,
-      "loss": 0.1955,
+      "loss": 0.196,
       "step": 8000
     },
     {
       "epoch": 1.4738393515106853,
-      "eval_loss": 0.9722194671630859,
-      "eval_runtime": 74.4255,
-      "eval_samples_per_second": 16.218,
-      "eval_steps_per_second": 2.029,
+      "eval_loss": 0.9711655378341675,
+      "eval_runtime": 74.3537,
+      "eval_samples_per_second": 16.233,
+      "eval_steps_per_second": 2.031,
       "step": 8000
     },
     {
       "epoch": 1.565954310980103,
-      "grad_norm": 43.05193328857422,
+      "grad_norm": 46.64162826538086,
       "learning_rate": 3.042557111274871e-08,
-      "loss": 0.1943,
+      "loss": 0.1946,
       "step": 8500
     },
     {
       "epoch": 1.565954310980103,
-      "eval_loss": 0.962623119354248,
-      "eval_runtime": 74.5615,
-      "eval_samples_per_second": 16.188,
-      "eval_steps_per_second": 2.025,
+      "eval_loss": 0.9621181488037109,
+      "eval_runtime": 75.7205,
+      "eval_samples_per_second": 15.94,
+      "eval_steps_per_second": 1.994,
       "step": 8500
     },
     {
       "epoch": 1.658069270449521,
-      "grad_norm": 0.050531528890132904,
+      "grad_norm": 0.05690698325634003,
       "learning_rate": 2.927413411938099e-08,
-      "loss": 0.2421,
+      "loss": 0.2422,
       "step": 9000
     },
     {
       "epoch": 1.658069270449521,
-      "eval_loss": 0.9543817043304443,
-      "eval_runtime": 76.2957,
-      "eval_samples_per_second": 15.82,
-      "eval_steps_per_second": 1.979,
+      "eval_loss": 0.9536784291267395,
+      "eval_runtime": 75.1745,
+      "eval_samples_per_second": 16.056,
+      "eval_steps_per_second": 2.009,
       "step": 9000
     },
     {
       "epoch": 1.750184229918939,
-      "grad_norm": 0.07409189641475677,
+      "grad_norm": 0.08470064401626587,
       "learning_rate": 2.8122697126013263e-08,
-      "loss": 0.2109,
+      "loss": 0.2106,
       "step": 9500
     },
     {
       "epoch": 1.750184229918939,
-      "eval_loss": 0.9468172192573547,
-      "eval_runtime": 74.551,
-      "eval_samples_per_second": 16.19,
-      "eval_steps_per_second": 2.025,
+      "eval_loss": 0.9457550644874573,
+      "eval_runtime": 74.5062,
+      "eval_samples_per_second": 16.2,
+      "eval_steps_per_second": 2.027,
       "step": 9500
     },
     {
       "epoch": 1.8422991893883567,
-      "grad_norm": 1.4242560863494873,
+      "grad_norm": 1.5745967626571655,
       "learning_rate": 2.697126013264554e-08,
-      "loss": 0.1799,
+      "loss": 0.1801,
       "step": 10000
     },
     {
       "epoch": 1.8422991893883567,
-      "eval_loss": 0.9402961730957031,
-      "eval_runtime": 74.4464,
-      "eval_samples_per_second": 16.213,
-      "eval_steps_per_second": 2.028,
+      "eval_loss": 0.9393072724342346,
+      "eval_runtime": 75.9432,
+      "eval_samples_per_second": 15.893,
+      "eval_steps_per_second": 1.988,
       "step": 10000
     },
     {
       "epoch": 1.9344141488577744,
-      "grad_norm": 0.06974973529577255,
+      "grad_norm": 0.06661002337932587,
       "learning_rate": 2.5819823139277818e-08,
-      "loss": 0.2114,
+      "loss": 0.2117,
       "step": 10500
     },
     {
       "epoch": 1.9344141488577744,
-      "eval_loss": 0.9320051074028015,
-      "eval_runtime": 76.0983,
-      "eval_samples_per_second": 15.861,
-      "eval_steps_per_second": 1.984,
+      "eval_loss": 0.9308408498764038,
+      "eval_runtime": 75.0811,
+      "eval_samples_per_second": 16.076,
+      "eval_steps_per_second": 2.011,
       "step": 10500
     },
     {
       "epoch": 2.026529108327192,
-      "grad_norm": 0.17525330185890198,
+      "grad_norm": 0.236691415309906,
       "learning_rate": 2.4668386145910094e-08,
-      "loss": 0.2058,
+      "loss": 0.2061,
       "step": 11000
     },
     {
       "epoch": 2.026529108327192,
-      "eval_loss": 0.9245836138725281,
-      "eval_runtime": 74.5035,
-      "eval_samples_per_second": 16.201,
-      "eval_steps_per_second": 2.027,
+      "eval_loss": 0.9237484335899353,
+      "eval_runtime": 74.6134,
+      "eval_samples_per_second": 16.177,
+      "eval_steps_per_second": 2.024,
       "step": 11000
     },
     {
       "epoch": 2.1186440677966103,
-      "grad_norm": 9.384895324707031,
+      "grad_norm": 8.831416130065918,
       "learning_rate": 2.3516949152542374e-08,
-      "loss": 0.188,
+      "loss": 0.1878,
       "step": 11500
     },
     {
       "epoch": 2.1186440677966103,
-      "eval_loss": 0.9178166389465332,
-      "eval_runtime": 74.4204,
-      "eval_samples_per_second": 16.219,
-      "eval_steps_per_second": 2.029,
+      "eval_loss": 0.9166750311851501,
+      "eval_runtime": 75.6886,
+      "eval_samples_per_second": 15.947,
+      "eval_steps_per_second": 1.995,
       "step": 11500
     },
     {
       "epoch": 2.210759027266028,
-      "grad_norm": 0.770260214805603,
+      "grad_norm": 0.787421464920044,
       "learning_rate": 2.236551215917465e-08,
-      "loss": 0.1657,
+      "loss": 0.1655,
       "step": 12000
     },
     {
       "epoch": 2.210759027266028,
-      "eval_loss": 0.9122205376625061,
-      "eval_runtime": 76.1185,
-      "eval_samples_per_second": 15.857,
-      "eval_steps_per_second": 1.984,
+      "eval_loss": 0.9109458327293396,
+      "eval_runtime": 74.9866,
+      "eval_samples_per_second": 16.096,
+      "eval_steps_per_second": 2.014,
       "step": 12000
     },
     {
       "epoch": 2.3028739867354457,
-      "grad_norm": 614.7385864257812,
+      "grad_norm": 535.4310913085938,
       "learning_rate": 2.1214075165806926e-08,
-      "loss": 0.1948,
+      "loss": 0.1946,
       "step": 12500
     },
     {
       "epoch": 2.3028739867354457,
-      "eval_loss": 0.90776127576828,
-      "eval_runtime": 74.5589,
-      "eval_samples_per_second": 16.189,
-      "eval_steps_per_second": 2.025,
+      "eval_loss": 0.9070548415184021,
+      "eval_runtime": 74.3796,
+      "eval_samples_per_second": 16.228,
+      "eval_steps_per_second": 2.03,
       "step": 12500
     },
     {
       "epoch": 2.3949889462048635,
-      "grad_norm": 0.7871919274330139,
+      "grad_norm": 0.8420014381408691,
       "learning_rate": 2.0062638172439202e-08,
       "loss": 0.1882,
       "step": 13000
     },
     {
       "epoch": 2.3949889462048635,
-      "eval_loss": 0.9029939770698547,
-      "eval_runtime": 74.3684,
-      "eval_samples_per_second": 16.23,
-      "eval_steps_per_second": 2.03,
+      "eval_loss": 0.9021016359329224,
+      "eval_runtime": 75.8812,
+      "eval_samples_per_second": 15.906,
+      "eval_steps_per_second": 1.99,
       "step": 13000
     },
     {
       "epoch": 2.4871039056742816,
-      "grad_norm": 1.3432903289794922,
+      "grad_norm": 1.4041686058044434,
       "learning_rate": 1.891120117907148e-08,
-      "loss": 0.1874,
+      "loss": 0.1871,
       "step": 13500
     },
     {
       "epoch": 2.4871039056742816,
-      "eval_loss": 0.8971672654151917,
-      "eval_runtime": 76.1444,
-      "eval_samples_per_second": 15.851,
-      "eval_steps_per_second": 1.983,
+      "eval_loss": 0.8960274457931519,
+      "eval_runtime": 75.2283,
+      "eval_samples_per_second": 16.044,
+      "eval_steps_per_second": 2.007,
       "step": 13500
     },
     {
       "epoch": 2.5792188651436994,
-      "grad_norm": 343.2808532714844,
+      "grad_norm": 327.2938537597656,
       "learning_rate": 1.7759764185703757e-08,
-      "loss": 0.1423,
+      "loss": 0.1419,
       "step": 14000
     },
     {
       "epoch": 2.5792188651436994,
-      "eval_loss": 0.8923307657241821,
-      "eval_runtime": 74.6823,
-      "eval_samples_per_second": 16.162,
-      "eval_steps_per_second": 2.022,
+      "eval_loss": 0.8913142085075378,
+      "eval_runtime": 74.4503,
+      "eval_samples_per_second": 16.212,
+      "eval_steps_per_second": 2.028,
       "step": 14000
     },
     {
       "epoch": 2.671333824613117,
-      "grad_norm": 0.01757567748427391,
+      "grad_norm": 0.01779676228761673,
       "learning_rate": 1.6608327192336037e-08,
-      "loss": 0.1424,
+      "loss": 0.1431,
       "step": 14500
     },
     {
       "epoch": 2.671333824613117,
-      "eval_loss": 0.888923704624176,
-      "eval_runtime": 74.5202,
-      "eval_samples_per_second": 16.197,
-      "eval_steps_per_second": 2.026,
+      "eval_loss": 0.8879275321960449,
+      "eval_runtime": 75.8636,
+      "eval_samples_per_second": 15.91,
+      "eval_steps_per_second": 1.99,
       "step": 14500
     },
     {
       "epoch": 2.7634487840825352,
-      "grad_norm": 1.6630617380142212,
+      "grad_norm": 1.6428073644638062,
       "learning_rate": 1.5456890198968313e-08,
-      "loss": 0.18,
+      "loss": 0.1811,
       "step": 15000
     },
     {
       "epoch": 2.7634487840825352,
-      "eval_loss": 0.885770320892334,
-      "eval_runtime": 76.123,
-      "eval_samples_per_second": 15.856,
-      "eval_steps_per_second": 1.984,
+      "eval_loss": 0.8848333954811096,
+      "eval_runtime": 75.1911,
+      "eval_samples_per_second": 16.052,
+      "eval_steps_per_second": 2.008,
       "step": 15000
     },
     {
       "epoch": 2.855563743551953,
-      "grad_norm": 0.0005675083957612514,
+      "grad_norm": 0.0006026208866387606,
       "learning_rate": 1.430545320560059e-08,
-      "loss": 0.1699,
+      "loss": 0.1694,
       "step": 15500
     },
     {
       "epoch": 2.855563743551953,
-      "eval_loss": 0.8834386467933655,
-      "eval_runtime": 74.4405,
-      "eval_samples_per_second": 16.214,
-      "eval_steps_per_second": 2.028,
+      "eval_loss": 0.8826532959938049,
+      "eval_runtime": 74.6044,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.024,
       "step": 15500
     }
   ],

checkpoint-15500/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:595bb6f39a89ec7030e395471c015665757cb5cacaafa6bba1613fd32b629686
+oid sha256:9a29c2374e59ce3f57b34db9cb20d2d1bdc870c3dad128e9a4d462636c37e8b0
 size 4847