Training in progress, epoch 0

README.md

@@ -30,20 +30,18 @@ should probably proofread and complete it, then remove this comment. -->
 
 # architectural_styles_classifier
 
-This model is a fine-tuned version of [nvidia/mit-b0](https://huggingface.co/nvidia/mit-b0) on the Architectural style dataset, retrieved from https://www.kaggle.com/datasets/dumitrux/architectural-styles-dataset.
+This model is a fine-tuned version of [nvidia/mit-b0](https://huggingface.co/nvidia/mit-b0) on the imagefolder dataset.
 It achieves the following results on the evaluation set:
 - Loss: 1.0414
 - Accuracy: 0.7252
 
 ## Model description
 
-Presentation link: https://www.canva.com/design/DAGLBMAs1K4/d8qvLN2nchSYVmnrwYzx0w/edit?utm_content=DAGLBMAs1K4&utm_campaign=designshare&utm_medium=link2&utm_source=sharebutton
-
-You can try the model inference in this link: https://huggingface.co/spaces/hanslab37/technospire
+More information needed
 
 ## Intended uses & limitations
 
-This model were developed only as part of personal experiment and portfolio to learn about developing image classification model with Huggingface. For now, it can be used to experiment only. Not recommended for daily use.
+More information needed
 
 ## Training and evaluation data
 

config.json

@@ -1,29 +1,32 @@
 {
-  "_name_or_path": "google/efficientnet-b3",
+  "_name_or_path": "nvidia/mit-b0",
   "architectures": [
-    "EfficientNetForImageClassification"
+    "SegformerForImageClassification"
   ],
-  "batch_norm_eps": 0.001,
-  "batch_norm_momentum": 0.99,
-  "depth_coefficient": 1.4,
-  "depth_divisor": 8,
-  "depthwise_padding": [
-    5,
-    18
+  "attention_probs_dropout_prob": 0.0,
+  "classifier_dropout_prob": 0.1,
+  "decoder_hidden_size": 256,
+  "depths": [
+    2,
+    2,
+    2,
+    2
   ],
-  "drop_connect_rate": 0.2,
-  "dropout_rate": 0.3,
-  "expand_ratios": [
+  "downsampling_rates": [
     1,
-    6,
-    6,
-    6,
-    6,
-    6,
-    6
+    4,
+    8,
+    16
+  ],
+  "drop_path_rate": 0.1,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.0,
+  "hidden_sizes": [
+    32,
+    64,
+    160,
+    256
   ],
-  "hidden_act": "swish",
-  "hidden_dim": 1536,
   "id2label": {
     "0": "Achaemenid architecture",
     "1": "American Foursquare architecture",
@@ -51,26 +54,8 @@
     "8": "Beaux-Arts architecture",
     "9": "Byzantine architecture"
   },
-  "image_size": 300,
-  "in_channels": [
-    32,
-    16,
-    24,
-    40,
-    80,
-    112,
-    192
-  ],
+  "image_size": 224,
   "initializer_range": 0.02,
-  "kernel_sizes": [
-    3,
-    3,
-    5,
-    3,
-    5,
-    5,
-    3
-  ],
   "label2id": {
     "Achaemenid architecture": "0",
     "American Foursquare architecture": "1",
@@ -98,40 +83,43 @@
     "Russian Revival architecture": "23",
     "Tudor Revival architecture": "24"
   },
-  "model_type": "efficientnet",
-  "num_block_repeats": [
+  "layer_norm_eps": 1e-06,
+  "mlp_ratios": [
+    4,
+    4,
+    4,
+    4
+  ],
+  "model_type": "segformer",
+  "num_attention_heads": [
     1,
     2,
-    2,
+    5,
+    8
+  ],
+  "num_channels": 3,
+  "num_encoder_blocks": 4,
+  "patch_sizes": [
+    7,
     3,
     3,
+    3
+  ],
+  "problem_type": "single_label_classification",
+  "reshape_last_stage": true,
+  "semantic_loss_ignore_index": 255,
+  "sr_ratios": [
+    8,
     4,
+    2,
     1
   ],
-  "num_channels": 3,
-  "num_hidden_layers": 64,
-  "out_channels": [
-    16,
-    24,
-    40,
-    80,
-    112,
-    192,
-    320
-  ],
-  "pooling_type": "mean",
-  "problem_type": "single_label_classification",
-  "squeeze_expansion_ratio": 0.25,
   "strides": [
-    1,
-    2,
-    2,
+    4,
     2,
-    1,
     2,
-    1
+    2
   ],
   "torch_dtype": "float32",
-  "transformers_version": "4.41.1",
-  "width_coefficient": 1.2
+  "transformers_version": "4.41.1"
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6e6c2eaaa25e67aaa55cede1bbdf928b6f85cc88902dbb7fe1450a4fd0a0bdab
-size 43362788
+oid sha256:26f7f2c0acbbf16f27ac01e53966d03fa366e8b3d3bd099165bd0558471dd627
+size 13325828

preprocessor_config.json

@@ -1,28 +1,22 @@
 {
   "_valid_processor_keys": [
     "images",
+    "segmentation_maps",
     "do_resize",
     "size",
     "resample",
-    "do_center_crop",
-    "crop_size",
     "do_rescale",
     "rescale_factor",
-    "rescale_offset",
     "do_normalize",
     "image_mean",
     "image_std",
-    "include_top",
+    "do_reduce_labels",
     "return_tensors",
     "data_format",
     "input_data_format"
   ],
-  "crop_size": {
-    "height": 289,
-    "width": 289
-  },
-  "do_center_crop": false,
   "do_normalize": true,
+  "do_reduce_labels": false,
   "do_rescale": true,
   "do_resize": true,
   "image_mean": [
@@ -30,18 +24,16 @@
     0.456,
     0.406
   ],
-  "image_processor_type": "EfficientNetImageProcessor",
+  "image_processor_type": "SegformerImageProcessor",
   "image_std": [
-    0.47853944,
-    0.4732864,
-    0.47434163
+    0.229,
+    0.224,
+    0.225
   ],
-  "include_top": true,
-  "resample": 0,
+  "resample": 2,
   "rescale_factor": 0.00392156862745098,
-  "rescale_offset": false,
   "size": {
-    "height": 300,
-    "width": 300
+    "height": 512,
+    "width": 512
   }
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a0f75eb877dbf4cd27fef65056fb69b6d89c9112a693d5d60f456b0ac42fbe15
+oid sha256:bdf7c5c1182d58471883402b8933ffb4a7a19706bf9b3b2d2557f4466c986f5b
 size 5112