Model save

README.md

@@ -1,12 +1,10 @@
 ---
 license: apache-2.0
-base_model: microsoft/beit-base-patch16-224-pt22k-ft22k
+base_model: google/vit-base-patch16-224-in21k
 tags:
-- image-classification
-- vision
 - generated_from_trainer
 metrics:
-- f1
+- accuracy
 model-index:
 - name: donut-base-beans
   results: []
@@ -17,10 +15,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # donut-base-beans
 
-This model is a fine-tuned version of [microsoft/beit-base-patch16-224-pt22k-ft22k](https://huggingface.co/microsoft/beit-base-patch16-224-pt22k-ft22k) on an unknown dataset.
+This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.0517
-- F1: 0.0
+- Accuracy: 0.4973
+- Loss: 0.0548
 
 ## Model description
 
@@ -45,17 +43,22 @@ The following hyperparameters were used during training:
 - seed: 1337
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- num_epochs: 5.0
+- num_epochs: 10.0
 
 ### Training results
 
-| Training Loss | Epoch | Step  | Validation Loss | F1  |
-|:-------------:|:-----:|:-----:|:---------------:|:---:|
-| 0.0672        | 1.0   | 13466 | 0.0696          | 0.0 |
-| 0.0612        | 2.0   | 26932 | 0.0588          | 0.0 |
-| 0.0496        | 3.0   | 40398 | 0.0556          | 0.0 |
-| 0.0443        | 4.0   | 53864 | 0.0535          | 0.0 |
-| 0.0683        | 5.0   | 67330 | 0.0517          | 0.0 |
+| Training Loss | Epoch | Step  | Accuracy | Validation Loss |
+|:-------------:|:-----:|:-----:|:--------:|:---------------:|
+| 0.0769        | 1.0   | 6733  | 0.4973   | 0.0657          |
+| 0.0729        | 2.0   | 13466 | 0.4973   | 0.0613          |
+| 0.0564        | 3.0   | 20199 | 0.4973   | 0.0621          |
+| 0.0534        | 4.0   | 26932 | 0.4973   | 0.0572          |
+| 0.0511        | 5.0   | 33665 | 0.4973   | 0.0562          |
+| 0.0659        | 6.0   | 40398 | 0.4973   | 0.0577          |
+| 0.0623        | 7.0   | 47131 | 0.4973   | 0.0567          |
+| 0.0575        | 8.0   | 53864 | 0.4973   | 0.0555          |
+| 0.048         | 9.0   | 60597 | 0.4973   | 0.0548          |
+| 0.0389        | 10.0  | 67330 | 0.4973   | 0.0548          |
 
 
 ### Framework versions

all_results.json

@@ -1,13 +1,14 @@
 {
-    "epoch": 5.0,
+    "epoch": 10.0,
+    "eval_accuracy": 0.49727767695099817,
     "eval_f1": 0.0,
-    "eval_loss": 0.05167969688773155,
-    "eval_runtime": 293.8942,
-    "eval_samples_per_second": 129.363,
-    "eval_steps_per_second": 8.088,
-    "total_flos": 8.34383324330106e+19,
-    "train_loss": 0.0714715420367653,
-    "train_runtime": 35752.3153,
-    "train_samples_per_second": 30.13,
-    "train_steps_per_second": 1.883
+    "eval_loss": 0.05480470508337021,
+    "eval_runtime": 903.7138,
+    "eval_samples_per_second": 42.07,
+    "eval_steps_per_second": 2.63,
+    "total_flos": 1.6694802102985576e+20,
+    "train_loss": 0.0,
+    "train_runtime": 0.0887,
+    "train_samples_per_second": 24287861.998,
+    "train_steps_per_second": 759048.532
 }

command.txt

@@ -1,7 +1,7 @@
 python run_image_classification.py \
     --train_dir /trunk/shared/eebo_data/images_cropped/ \
     --validation_dir /trunk/shared/eebo_data/images_cropped/ \
-    --output_dir ./microsoft_beit-base-patch16-224-pt22k-ft22k/ \
+    --output_dir ./default_model/ \
     --remove_unused_columns False \
     --label_column_name labels \
     --do_train \
@@ -10,8 +10,8 @@ python run_image_classification.py \
     --push_to_hub_model_id donut-base-beans \
     --learning_rate 3e-5 \
     --num_train_epochs 10 \
-    --per_device_train_batch_size 8 \
-    --per_device_eval_batch_size 8 \
+    --per_device_train_batch_size 64 \
+    --per_device_eval_batch_size 64 \
     --logging_strategy steps \
     --logging_steps 10 \
     --eval_strategy epoch \
@@ -22,5 +22,4 @@ python run_image_classification.py \
     --ignore_mismatched_sizes \
     --image_column_name image_url \
     --label_column_name label \
-    --model_name_or_path microsoft/beit-base-patch16-224-pt22k-ft22k \
-    --resume_from_checkpoint /trunk2/yufei/summer24/transformers/examples/pytorch/image-classification/microsoft_beit-base-patch16-224-pt22k-ft22k/checkpoint-33670
+    --resume_from_checkpoint /trunk2/yufei/summer24/transformers/examples/pytorch/image-classification/default_model/checkpoint-67330

config.json

@@ -1,15 +1,10 @@
 {
-  "_name_or_path": "microsoft/beit-base-patch16-224-pt22k-ft22k",
-  "add_fpn": false,
+  "_name_or_path": "google/vit-base-patch16-224-in21k",
   "architectures": [
-    "BeitForImageClassification"
+    "ViTForImageClassification"
   ],
   "attention_probs_dropout_prob": 0.0,
-  "auxiliary_channels": 256,
-  "auxiliary_concat_input": false,
-  "auxiliary_loss_weight": 0.4,
-  "auxiliary_num_convs": 1,
-  "drop_path_rate": 0.1,
+  "encoder_stride": 16,
   "finetuning_task": "image-classification",
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.0,
@@ -24,49 +19,12 @@
     "Value(dtype='int64', id=None)": "0"
   },
   "layer_norm_eps": 1e-12,
-  "layer_scale_init_value": 0.1,
-  "model_type": "beit",
+  "model_type": "vit",
   "num_attention_heads": 12,
   "num_channels": 3,
   "num_hidden_layers": 12,
-  "out_features": [
-    "stage12"
-  ],
-  "out_indices": [
-    12
-  ],
   "patch_size": 16,
-  "pool_scales": [
-    1,
-    2,
-    3,
-    6
-  ],
-  "problem_type": "regression",
-  "reshape_hidden_states": true,
-  "semantic_loss_ignore_index": 255,
-  "stage_names": [
-    "stem",
-    "stage1",
-    "stage2",
-    "stage3",
-    "stage4",
-    "stage5",
-    "stage6",
-    "stage7",
-    "stage8",
-    "stage9",
-    "stage10",
-    "stage11",
-    "stage12"
-  ],
+  "qkv_bias": true,
   "torch_dtype": "float32",
-  "transformers_version": "4.43.0.dev0",
-  "use_absolute_position_embeddings": false,
-  "use_auxiliary_head": true,
-  "use_mask_token": false,
-  "use_mean_pooling": true,
-  "use_relative_position_bias": true,
-  "use_shared_relative_position_bias": false,
-  "vocab_size": 8192
+  "transformers_version": "4.43.0.dev0"
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 5.0,
+    "epoch": 10.0,
     "eval_f1": 0.0,
-    "eval_loss": 0.05167969688773155,
-    "eval_runtime": 293.8942,
-    "eval_samples_per_second": 129.363,
-    "eval_steps_per_second": 8.088
+    "eval_loss": 0.05480470508337021,
+    "eval_runtime": 903.7138,
+    "eval_samples_per_second": 42.07,
+    "eval_steps_per_second": 2.63
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:87a3e53f87c26251c9485be8d2e6b4896fb3ca5fbc1fe742353ffe4c57e091e8
-size 343077252
+oid sha256:2561cfd7419a353fba1d1368529d633d6fb11d901bddcfe413ee3478cec81425
+size 343220892

preprocessor_config.json

@@ -1,11 +1,5 @@
 {
-  "crop_size": {
-    "height": 224,
-    "width": 224
-  },
-  "do_center_crop": false,
   "do_normalize": true,
-  "do_reduce_labels": false,
   "do_rescale": true,
   "do_resize": true,
   "image_mean": [
@@ -13,7 +7,7 @@
     0.5,
     0.5
   ],
-  "image_processor_type": "BeitImageProcessor",
+  "image_processor_type": "ViTImageProcessor",
   "image_std": [
     0.5,
     0.5,

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 5.0,
-    "total_flos": 8.34383324330106e+19,
-    "train_loss": 0.0714715420367653,
-    "train_runtime": 35752.3153,
-    "train_samples_per_second": 30.13,
-    "train_steps_per_second": 1.883
+    "epoch": 10.0,
+    "total_flos": 1.6694802102985576e+20,
+    "train_loss": 0.0,
+    "train_runtime": 0.0887,
+    "train_samples_per_second": 24287861.998,
+    "train_steps_per_second": 759048.532
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b437f98ee482d0b8115a8072d28d365612b268b96157a462218455ae7ca2971b
+oid sha256:7d8a6f533c72550417f1e253380ae12ad72153532482192012f764515f190bbf
 size 5304