End of training

README.md

@@ -14,10 +14,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # serverless-roomsort
 
-This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on the None dataset.
+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 the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.5517
-- Accuracy: 0.4380
+- Loss: 0.0394
+- Accuracy: 0.9892
 
 ## Model description
 
@@ -37,19 +37,23 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 2e-05
-- train_batch_size: 32
+- train_batch_size: 16
 - eval_batch_size: 64
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - lr_scheduler_warmup_steps: 500
-- num_epochs: 1
+- num_epochs: 5
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss | Accuracy |
 |:-------------:|:-----:|:----:|:---------------:|:--------:|
-| No log        | 1.0   | 77   | 1.5517          | 0.4380   |
+| 0.7844        | 1.0   | 762  | 0.0608          | 0.9791   |
+| 0.0361        | 2.0   | 1524 | 0.0626          | 0.9830   |
+| 0.0149        | 3.0   | 2286 | 0.0468          | 0.9879   |
+| 0.0027        | 4.0   | 3048 | 0.0394          | 0.9892   |
+| 0.0017        | 5.0   | 3810 | 0.0486          | 0.9889   |
 
 
 ### Framework versions

config.json

@@ -1,38 +1,72 @@
 {
-  "_name_or_path": "google/vit-base-patch16-224-in21k",
+  "_name_or_path": "microsoft/beit-base-patch16-224-pt22k-ft22k",
   "architectures": [
-    "ViTForImageClassification"
+    "BeitForImageClassification"
   ],
   "attention_probs_dropout_prob": 0.0,
-  "encoder_stride": 16,
+  "auxiliary_channels": 256,
+  "auxiliary_concat_input": false,
+  "auxiliary_loss_weight": 0.4,
+  "auxiliary_num_convs": 1,
+  "drop_path_rate": 0.1,
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.0,
   "hidden_size": 768,
   "id2label": {
     "0": "Bathroom",
     "1": "Kitchen",
-    "2": "Exterior",
-    "3": "Bedroom_Office",
-    "4": "LivingRoom_DiningRoom"
+    "2": "Bedroom",
+    "3": "LivingRoom_DiningRoom",
+    "4": "Garage",
+    "5": "Hallway",
+    "6": "LaundryRoom",
+    "7": "Office",
+    "8": "Blueprints",
+    "9": "Exterior"
   },
   "image_size": 224,
   "initializer_range": 0.02,
   "intermediate_size": 3072,
   "label2id": {
     "Bathroom": 0,
-    "Bedroom_Office": 3,
-    "Exterior": 2,
+    "Bedroom": 2,
+    "Blueprints": 8,
+    "Exterior": 9,
+    "Garage": 4,
+    "Hallway": 5,
     "Kitchen": 1,
-    "LivingRoom_DiningRoom": 4
+    "LaundryRoom": 6,
+    "LivingRoom_DiningRoom": 3,
+    "Office": 7
   },
   "layer_norm_eps": 1e-12,
-  "model_type": "vit",
+  "layer_scale_init_value": 0.1,
+  "model_type": "beit",
   "num_attention_heads": 12,
   "num_channels": 3,
   "num_hidden_layers": 12,
+  "out_indices": [
+    3,
+    5,
+    7,
+    11
+  ],
   "patch_size": 16,
+  "pool_scales": [
+    1,
+    2,
+    3,
+    6
+  ],
   "problem_type": "single_label_classification",
-  "qkv_bias": true,
+  "semantic_loss_ignore_index": 255,
   "torch_dtype": "float32",
-  "transformers_version": "4.17.0"
+  "transformers_version": "4.17.0",
+  "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
 }

eval_results.txt

@@ -1,6 +1,6 @@
-epoch = 1.0
-eval_accuracy = 0.43795620437956206
-eval_loss = 1.5516855716705322
-eval_runtime = 6.2916
-eval_samples_per_second = 43.55
-eval_steps_per_second = 0.795
+epoch = 5.0
+eval_accuracy = 0.9891625615763546
+eval_loss = 0.03943309187889099
+eval_runtime = 76.6659
+eval_samples_per_second = 52.957
+eval_steps_per_second = 0.835

preprocessor_config.json

@@ -1,7 +1,9 @@
 {
+  "crop_size": 224,
+  "do_center_crop": false,
   "do_normalize": true,
   "do_resize": true,
-  "feature_extractor_type": "ViTFeatureExtractor",
+  "feature_extractor_type": "BeitFeatureExtractor",
   "image_mean": [
     0.5,
     0.5,
@@ -12,6 +14,7 @@
     0.5,
     0.5
   ],
+  "reduce_labels": false,
   "resample": 2,
   "size": 224
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d587760ba20be9578d4768849aeba99d441a3946f2774570c3b1ec6dee66cc60
-size 343289969
+oid sha256:f2c635720c1af94caa1e0bff1923a0b882a153afc5d07da2247d3b5691f82f84
+size 346898795

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:628c1cdb8971b26ff1e3e30174de92aa6d4303e97c0acf3281e63ad2bae7cdbe
+oid sha256:c792df19980ada2f81b6c845354cf210f3968795289de29329e37f932985207b
 size 3055