End of training

README.md

@@ -1,8 +1,9 @@
 ---
+base_model: ./clip-roberta
 tags:
 - generated_from_trainer
 datasets:
-- coco_dataset_script
+- ydshieh/coco_dataset_script
 model-index:
 - name: clip-roberta-finetuned
   results: []
@@ -13,7 +14,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # clip-roberta-finetuned
 
-This model was trained from scratch on the coco_dataset_script dataset.
+This model is a fine-tuned version of [./clip-roberta](https://huggingface.co/./clip-roberta) on the ydshieh/coco_dataset_script 2017 dataset.
+It achieves the following results on the evaluation set:
+- Loss: 1.9692
 
 ## Model description
 

all_results.json

@@ -1,11 +1,11 @@
 {
     "epoch": 3.0,
-    "eval_loss": 1.696897029876709,
-    "eval_runtime": 305.6648,
-    "eval_samples_per_second": 81.835,
-    "eval_steps_per_second": 1.279,
-    "train_loss": 0.37869084192067465,
-    "train_runtime": 53685.7584,
-    "train_samples_per_second": 33.068,
-    "train_steps_per_second": 0.517
+    "eval_loss": 1.9691866636276245,
+    "eval_runtime": 205.5006,
+    "eval_samples_per_second": 121.722,
+    "eval_steps_per_second": 0.954,
+    "train_loss": 0.481756305474586,
+    "train_runtime": 278221.0505,
+    "train_samples_per_second": 6.381,
+    "train_steps_per_second": 0.05
 }

eval_results.json

@@ -1,7 +1,7 @@
 {
     "epoch": 3.0,
-    "eval_loss": 1.696897029876709,
-    "eval_runtime": 305.6648,
-    "eval_samples_per_second": 81.835,
-    "eval_steps_per_second": 1.279
+    "eval_loss": 1.9691866636276245,
+    "eval_runtime": 205.5006,
+    "eval_samples_per_second": 121.722,
+    "eval_steps_per_second": 0.954
 }

train_results.json

@@ -1,7 +1,7 @@
 {
     "epoch": 3.0,
-    "train_loss": 0.37869084192067465,
-    "train_runtime": 53685.7584,
-    "train_samples_per_second": 33.068,
-    "train_steps_per_second": 0.517
+    "train_loss": 0.481756305474586,
+    "train_runtime": 278221.0505,
+    "train_samples_per_second": 6.381,
+    "train_steps_per_second": 0.05
 }

trainer_state.json

@@ -3,353 +3,185 @@
   "best_model_checkpoint": null,
   "epoch": 3.0,
   "eval_steps": 500,
-  "global_step": 27741,
+  "global_step": 13872,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
-    {
-      "epoch": 0.05,
-      "learning_rate": 4.909880682022999e-05,
-      "loss": 1.3107,
-      "step": 500
-    },
     {
       "epoch": 0.11,
-      "learning_rate": 4.819761364045997e-05,
-      "loss": 0.9349,
-      "step": 1000
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 4.7296420460689956e-05,
-      "loss": 0.8565,
-      "step": 1500
+      "learning_rate": 4.819780853517878e-05,
+      "loss": 1.4429,
+      "step": 500
     },
     {
       "epoch": 0.22,
-      "learning_rate": 4.639522728091994e-05,
-      "loss": 0.7897,
-      "step": 2000
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 4.5494034101149925e-05,
-      "loss": 0.7405,
-      "step": 2500
+      "learning_rate": 4.639561707035756e-05,
+      "loss": 0.9764,
+      "step": 1000
     },
     {
       "epoch": 0.32,
-      "learning_rate": 4.459284092137991e-05,
-      "loss": 0.7032,
-      "step": 3000
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 4.3691647741609894e-05,
-      "loss": 0.6635,
-      "step": 3500
+      "learning_rate": 4.459342560553633e-05,
+      "loss": 0.8637,
+      "step": 1500
     },
     {
       "epoch": 0.43,
-      "learning_rate": 4.279045456183988e-05,
-      "loss": 0.6374,
-      "step": 4000
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 4.188926138206986e-05,
-      "loss": 0.6199,
-      "step": 4500
+      "learning_rate": 4.2791234140715114e-05,
+      "loss": 0.7835,
+      "step": 2000
     },
     {
       "epoch": 0.54,
-      "learning_rate": 4.098806820229985e-05,
-      "loss": 0.5862,
-      "step": 5000
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.008687502252983e-05,
-      "loss": 0.5686,
-      "step": 5500
+      "learning_rate": 4.098904267589389e-05,
+      "loss": 0.7204,
+      "step": 2500
     },
     {
       "epoch": 0.65,
-      "learning_rate": 3.9185681842759816e-05,
-      "loss": 0.5402,
-      "step": 6000
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 3.82844886629898e-05,
-      "loss": 0.5319,
-      "step": 6500
+      "learning_rate": 3.9186851211072664e-05,
+      "loss": 0.6731,
+      "step": 3000
     },
     {
       "epoch": 0.76,
-      "learning_rate": 3.7383295483219785e-05,
-      "loss": 0.5051,
-      "step": 7000
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 3.648210230344977e-05,
-      "loss": 0.4788,
-      "step": 7500
+      "learning_rate": 3.7384659746251445e-05,
+      "loss": 0.6355,
+      "step": 3500
     },
     {
       "epoch": 0.87,
-      "learning_rate": 3.558090912367975e-05,
-      "loss": 0.4726,
-      "step": 8000
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 3.467971594390974e-05,
-      "loss": 0.4583,
-      "step": 8500
+      "learning_rate": 3.558246828143022e-05,
+      "loss": 0.5884,
+      "step": 4000
     },
     {
       "epoch": 0.97,
-      "learning_rate": 3.3778522764139723e-05,
-      "loss": 0.4513,
-      "step": 9000
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 3.287732958436971e-05,
-      "loss": 0.4005,
-      "step": 9500
+      "learning_rate": 3.3780276816608994e-05,
+      "loss": 0.5598,
+      "step": 4500
     },
     {
       "epoch": 1.08,
-      "learning_rate": 3.197613640459969e-05,
-      "loss": 0.3658,
-      "step": 10000
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 3.107494322482968e-05,
-      "loss": 0.3613,
-      "step": 10500
+      "learning_rate": 3.1978085351787776e-05,
+      "loss": 0.4735,
+      "step": 5000
     },
     {
       "epoch": 1.19,
-      "learning_rate": 3.017375004505966e-05,
-      "loss": 0.3473,
-      "step": 11000
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 2.9272556865289646e-05,
-      "loss": 0.3433,
-      "step": 11500
+      "learning_rate": 3.0175893886966554e-05,
+      "loss": 0.4465,
+      "step": 5500
     },
     {
       "epoch": 1.3,
-      "learning_rate": 2.837136368551963e-05,
-      "loss": 0.3344,
-      "step": 12000
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 2.7470170505749615e-05,
-      "loss": 0.334,
-      "step": 12500
+      "learning_rate": 2.8373702422145332e-05,
+      "loss": 0.43,
+      "step": 6000
     },
     {
       "epoch": 1.41,
-      "learning_rate": 2.65689773259796e-05,
-      "loss": 0.315,
-      "step": 13000
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 2.5667784146209584e-05,
-      "loss": 0.3237,
-      "step": 13500
+      "learning_rate": 2.657151095732411e-05,
+      "loss": 0.4175,
+      "step": 6500
     },
     {
       "epoch": 1.51,
-      "learning_rate": 2.476659096643957e-05,
-      "loss": 0.3172,
-      "step": 14000
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 2.3865397786669553e-05,
-      "loss": 0.2937,
-      "step": 14500
+      "learning_rate": 2.4769319492502884e-05,
+      "loss": 0.4104,
+      "step": 7000
     },
     {
       "epoch": 1.62,
-      "learning_rate": 2.2964204606899537e-05,
-      "loss": 0.2964,
-      "step": 15000
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 2.2063011427129522e-05,
-      "loss": 0.2896,
-      "step": 15500
+      "learning_rate": 2.2967128027681662e-05,
+      "loss": 0.3902,
+      "step": 7500
     },
     {
       "epoch": 1.73,
-      "learning_rate": 2.1161818247359506e-05,
-      "loss": 0.2855,
-      "step": 16000
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 2.026062506758949e-05,
-      "loss": 0.2753,
-      "step": 16500
+      "learning_rate": 2.116493656286044e-05,
+      "loss": 0.3779,
+      "step": 8000
     },
     {
       "epoch": 1.84,
-      "learning_rate": 1.9359431887819472e-05,
-      "loss": 0.2691,
-      "step": 17000
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 1.8458238708049457e-05,
-      "loss": 0.272,
-      "step": 17500
+      "learning_rate": 1.936274509803922e-05,
+      "loss": 0.3634,
+      "step": 8500
     },
     {
       "epoch": 1.95,
-      "learning_rate": 1.755704552827944e-05,
-      "loss": 0.2653,
-      "step": 18000
-    },
-    {
-      "epoch": 2.0,
-      "learning_rate": 1.6655852348509426e-05,
-      "loss": 0.2495,
-      "step": 18500
+      "learning_rate": 1.7560553633217993e-05,
+      "loss": 0.3584,
+      "step": 9000
     },
     {
       "epoch": 2.05,
-      "learning_rate": 1.575465916873941e-05,
-      "loss": 0.2024,
-      "step": 19000
-    },
-    {
-      "epoch": 2.11,
-      "learning_rate": 1.4853465988969395e-05,
-      "loss": 0.1984,
-      "step": 19500
+      "learning_rate": 1.575836216839677e-05,
+      "loss": 0.3085,
+      "step": 9500
     },
     {
       "epoch": 2.16,
-      "learning_rate": 1.3952272809199379e-05,
-      "loss": 0.1957,
-      "step": 20000
-    },
-    {
-      "epoch": 2.22,
-      "learning_rate": 1.3051079629429367e-05,
-      "loss": 0.1935,
-      "step": 20500
+      "learning_rate": 1.395617070357555e-05,
+      "loss": 0.2635,
+      "step": 10000
     },
     {
       "epoch": 2.27,
-      "learning_rate": 1.214988644965935e-05,
-      "loss": 0.1878,
-      "step": 21000
-    },
-    {
-      "epoch": 2.33,
-      "learning_rate": 1.1248693269889334e-05,
-      "loss": 0.1815,
-      "step": 21500
+      "learning_rate": 1.2153979238754325e-05,
+      "loss": 0.2616,
+      "step": 10500
     },
     {
       "epoch": 2.38,
-      "learning_rate": 1.0347500090119317e-05,
-      "loss": 0.1855,
-      "step": 22000
-    },
-    {
-      "epoch": 2.43,
-      "learning_rate": 9.446306910349303e-06,
-      "loss": 0.1762,
-      "step": 22500
+      "learning_rate": 1.0351787773933102e-05,
+      "loss": 0.2543,
+      "step": 11000
     },
     {
       "epoch": 2.49,
-      "learning_rate": 8.545113730579288e-06,
-      "loss": 0.173,
-      "step": 23000
-    },
-    {
-      "epoch": 2.54,
-      "learning_rate": 7.643920550809272e-06,
-      "loss": 0.1717,
-      "step": 23500
+      "learning_rate": 8.54959630911188e-06,
+      "loss": 0.2477,
+      "step": 11500
     },
     {
       "epoch": 2.6,
-      "learning_rate": 6.742727371039257e-06,
-      "loss": 0.1703,
-      "step": 24000
-    },
-    {
-      "epoch": 2.65,
-      "learning_rate": 5.84153419126924e-06,
-      "loss": 0.1636,
-      "step": 24500
+      "learning_rate": 6.747404844290659e-06,
+      "loss": 0.2451,
+      "step": 12000
     },
     {
       "epoch": 2.7,
-      "learning_rate": 4.940341011499225e-06,
-      "loss": 0.1652,
-      "step": 25000
-    },
-    {
-      "epoch": 2.76,
-      "learning_rate": 4.03914783172921e-06,
-      "loss": 0.1609,
-      "step": 25500
+      "learning_rate": 4.945213379469435e-06,
+      "loss": 0.2399,
+      "step": 12500
     },
     {
       "epoch": 2.81,
-      "learning_rate": 3.1379546519591943e-06,
-      "loss": 0.1624,
-      "step": 26000
-    },
-    {
-      "epoch": 2.87,
-      "learning_rate": 2.2367614721891784e-06,
-      "loss": 0.1541,
-      "step": 26500
+      "learning_rate": 3.143021914648212e-06,
+      "loss": 0.2371,
+      "step": 13000
     },
     {
       "epoch": 2.92,
-      "learning_rate": 1.335568292419163e-06,
-      "loss": 0.1571,
-      "step": 27000
-    },
-    {
-      "epoch": 2.97,
-      "learning_rate": 4.3437511264914753e-07,
-      "loss": 0.1525,
-      "step": 27500
+      "learning_rate": 1.3408304498269898e-06,
+      "loss": 0.2296,
+      "step": 13500
     },
     {
       "epoch": 3.0,
-      "step": 27741,
+      "step": 13872,
       "total_flos": 2.3702775297552e+17,
-      "train_loss": 0.37869084192067465,
-      "train_runtime": 53685.7584,
-      "train_samples_per_second": 33.068,
-      "train_steps_per_second": 0.517
+      "train_loss": 0.481756305474586,
+      "train_runtime": 278221.0505,
+      "train_samples_per_second": 6.381,
+      "train_steps_per_second": 0.05
     }
   ],
   "logging_steps": 500,
-  "max_steps": 27741,
+  "max_steps": 13872,
   "num_train_epochs": 3,
   "save_steps": 500,
   "total_flos": 2.3702775297552e+17,