Hugging Face's logo

monai-test
/
pathology_nuclei_classification

monai
medical
Model card Files Community
katielink commited on
Commit
7cec656
•
1 Parent(s): c0b6224

Update deterministic results

Browse files
Files changed (6) hide show
  1. README.md +15 -15
  2. configs/metadata.json +3 -2
  3. configs/train.json +13 -2
  4. docs/README.md +15 -15
  5. models/model.pt +1 -1
  6. models/model.ts +2 -2
README.md CHANGED
@@ -106,40 +106,40 @@ The training was performed with the following:
106
  ## Performance
107
  This model achieves the following F1 score on the validation data provided as part of the dataset:
108
 
109
- - Train F1 score = 0.941
110
- - Validation F1 score = 0.840
111
 
112
  <hr/>
113
- Confusion Metrics for <b>Validation</b> for individual classes are (at epoch 50):
114
 
115
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
116
  |-----------|--------|--------------|------------|----------------|
117
- | Precision | 0.6250 | 0.7085 | 0.9188 | 0.8571 |
118
- | Recall | 0.1449 | 0.8750 | 0.9310 | 0.8154 |
119
- | F1-score | 0.2353 | 0.7830 | 0.9249 | 0.8357 |
120
 
121
 
122
  <hr/>
123
- Confusion Metrics for <b>Training</b> for individual classes are (at epoch 50):
124
 
125
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
126
  |-----------|--------|--------------|------------|----------------|
127
- | Precision | 0.8902 | 0.9418 | 0.9717 | 0.9189 |
128
- | Recall | 0.7935 | 0.9250 | 0.9725 | 0.9345 |
129
- | F1-score | 0.8391 | 0.9333 | 0.9721 | 0.9267 |
130
 
131
 
132
 
133
  #### Training Loss and F1
134
- A graph showing the training Loss and F1-score over 50 epochs.
135
 
136
- ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_train_loss_v2.png) <br>
137
- ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_train_f1_v2.png) <br>
138
 
139
  #### Validation F1
140
- A graph showing the validation F1-score over 50 epochs.
141
 
142
- ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_val_f1_v2.png) <br>
143
 
144
  ## MONAI Bundle Commands
145
  In addition to the Pythonic APIs, a few command line interfaces (CLI) are provided to interact with the bundle. The CLI supports flexible use cases, such as overriding configs at runtime and predefining arguments in a file.
 
106
  ## Performance
107
  This model achieves the following F1 score on the validation data provided as part of the dataset:
108
 
109
+ - Train F1 score = 0.926
110
+ - Validation F1 score = 0.852
111
 
112
  <hr/>
113
+ Confusion Metrics for <b>Validation</b> for individual classes are:
114
 
115
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
116
  |-----------|--------|--------------|------------|----------------|
117
+ | Precision | 0.6909 | 0.7773 | 0.9078 | 0.8478 |
118
+ | Recall | 0.2754 | 0.7831 | 0.9533 | 0.8514 |
119
+ | F1-score | 0.3938 | 0.7802 | 0.9300 | 0.8496 |
120
 
121
 
122
  <hr/>
123
+ Confusion Metrics for <b>Training</b> for individual classes are:
124
 
125
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
126
  |-----------|--------|--------------|------------|----------------|
127
+ | Precision | 0.8000 | 0.9076 | 0.9560 | 0.9019 |
128
+ | Recall | 0.6512 | 0.9028 | 0.9690 | 0.8989 |
129
+ | F1-score | 0.7179 | 0.9052 | 0.9625 | 0.9004 |
130
 
131
 
132
 
133
  #### Training Loss and F1
134
+ A graph showing the training Loss and F1-score over 100 epochs.
135
 
136
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_train_loss_v3.png) <br>
137
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_train_f1_v3.png) <br>
138
 
139
  #### Validation F1
140
+ A graph showing the validation F1-score over 100 epochs.
141
 
142
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_val_f1_v3.png) <br>
143
 
144
  ## MONAI Bundle Commands
145
  In addition to the Pythonic APIs, a few command line interfaces (CLI) are provided to interact with the bundle. The CLI supports flexible use cases, such as overriding configs at runtime and predefining arguments in a file.
configs/metadata.json CHANGED
@@ -1,7 +1,8 @@
1
  {
2
  "schema": "https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/meta_schema_20220324.json",
3
- "version": "0.0.9",
4
  "changelog": {
 
5
  "0.0.9": "Update README Formatting",
6
  "0.0.8": "enable deterministic training",
7
  "0.0.7": "update benchmark on A100",
@@ -31,7 +32,7 @@
31
  "label_classes": "single channel data",
32
  "pred_classes": "4 channels OneHot data, channel 0 is Other, channel 1 is Inflammatory, channel 2 is Epithelial, channel 3 is Spindle-Shaped",
33
  "eval_metrics": {
34
- "f1_score": 0.84
35
  },
36
  "intended_use": "This is an example, not to be used for diagnostic purposes",
37
  "references": [
 
1
  {
2
  "schema": "https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/meta_schema_20220324.json",
3
+ "version": "0.1.0",
4
  "changelog": {
5
+ "0.1.0": "Update deterministic results",
6
  "0.0.9": "Update README Formatting",
7
  "0.0.8": "enable deterministic training",
8
  "0.0.7": "update benchmark on A100",
 
32
  "label_classes": "single channel data",
33
  "pred_classes": "4 channels OneHot data, channel 0 is Other, channel 1 is Inflammatory, channel 2 is Epithelial, channel 3 is Spindle-Shaped",
34
  "eval_metrics": {
35
+ "f1_score": 0.852
36
  },
37
  "intended_use": "This is an example, not to be used for diagnostic purposes",
38
  "references": [
configs/train.json CHANGED
@@ -28,9 +28,15 @@
28
  "optimizer": {
29
  "_target_": "torch.optim.Adam",
30
  "params": "$@network.parameters()",
31
- "lr": 0.0001
 
32
  },
33
- "max_epochs": 50,
 
 
 
 
 
34
  "train": {
35
  "preprocessing": {
36
  "_target_": "Compose",
@@ -159,6 +165,11 @@
159
  ]
160
  },
161
  "handlers": [
 
 
 
 
 
162
  {
163
  "_target_": "ValidationHandler",
164
  "validator": "@validate#evaluator",
 
28
  "optimizer": {
29
  "_target_": "torch.optim.Adam",
30
  "params": "$@network.parameters()",
31
+ "lr": 0.0001,
32
+ "weight_decay": 1e-05
33
  },
34
+ "lr_scheduler": {
35
+ "_target_": "torch.optim.lr_scheduler.StepLR",
36
+ "optimizer": "@optimizer",
37
+ "step_size": 50
38
+ },
39
+ "max_epochs": 100,
40
  "train": {
41
  "preprocessing": {
42
  "_target_": "Compose",
 
165
  ]
166
  },
167
  "handlers": [
168
+ {
169
+ "_target_": "LrScheduleHandler",
170
+ "lr_scheduler": "@lr_scheduler",
171
+ "print_lr": true
172
+ },
173
  {
174
  "_target_": "ValidationHandler",
175
  "validator": "@validate#evaluator",
docs/README.md CHANGED
@@ -99,40 +99,40 @@ The training was performed with the following:
99
  ## Performance
100
  This model achieves the following F1 score on the validation data provided as part of the dataset:
101
 
102
- - Train F1 score = 0.941
103
- - Validation F1 score = 0.840
104
 
105
  <hr/>
106
- Confusion Metrics for <b>Validation</b> for individual classes are (at epoch 50):
107
 
108
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
109
  |-----------|--------|--------------|------------|----------------|
110
- | Precision | 0.6250 | 0.7085 | 0.9188 | 0.8571 |
111
- | Recall | 0.1449 | 0.8750 | 0.9310 | 0.8154 |
112
- | F1-score | 0.2353 | 0.7830 | 0.9249 | 0.8357 |
113
 
114
 
115
  <hr/>
116
- Confusion Metrics for <b>Training</b> for individual classes are (at epoch 50):
117
 
118
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
119
  |-----------|--------|--------------|------------|----------------|
120
- | Precision | 0.8902 | 0.9418 | 0.9717 | 0.9189 |
121
- | Recall | 0.7935 | 0.9250 | 0.9725 | 0.9345 |
122
- | F1-score | 0.8391 | 0.9333 | 0.9721 | 0.9267 |
123
 
124
 
125
 
126
  #### Training Loss and F1
127
- A graph showing the training Loss and F1-score over 50 epochs.
128
 
129
- ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_train_loss_v2.png) <br>
130
- ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_train_f1_v2.png) <br>
131
 
132
  #### Validation F1
133
- A graph showing the validation F1-score over 50 epochs.
134
 
135
- ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_val_f1_v2.png) <br>
136
 
137
  ## MONAI Bundle Commands
138
  In addition to the Pythonic APIs, a few command line interfaces (CLI) are provided to interact with the bundle. The CLI supports flexible use cases, such as overriding configs at runtime and predefining arguments in a file.
 
99
  ## Performance
100
  This model achieves the following F1 score on the validation data provided as part of the dataset:
101
 
102
+ - Train F1 score = 0.926
103
+ - Validation F1 score = 0.852
104
 
105
  <hr/>
106
+ Confusion Metrics for <b>Validation</b> for individual classes are:
107
 
108
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
109
  |-----------|--------|--------------|------------|----------------|
110
+ | Precision | 0.6909 | 0.7773 | 0.9078 | 0.8478 |
111
+ | Recall | 0.2754 | 0.7831 | 0.9533 | 0.8514 |
112
+ | F1-score | 0.3938 | 0.7802 | 0.9300 | 0.8496 |
113
 
114
 
115
  <hr/>
116
+ Confusion Metrics for <b>Training</b> for individual classes are:
117
 
118
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
119
  |-----------|--------|--------------|------------|----------------|
120
+ | Precision | 0.8000 | 0.9076 | 0.9560 | 0.9019 |
121
+ | Recall | 0.6512 | 0.9028 | 0.9690 | 0.8989 |
122
+ | F1-score | 0.7179 | 0.9052 | 0.9625 | 0.9004 |
123
 
124
 
125
 
126
  #### Training Loss and F1
127
+ A graph showing the training Loss and F1-score over 100 epochs.
128
 
129
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_train_loss_v3.png) <br>
130
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_train_f1_v3.png) <br>
131
 
132
  #### Validation F1
133
+ A graph showing the validation F1-score over 100 epochs.
134
 
135
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_val_f1_v3.png) <br>
136
 
137
  ## MONAI Bundle Commands
138
  In addition to the Pythonic APIs, a few command line interfaces (CLI) are provided to interact with the bundle. The CLI supports flexible use cases, such as overriding configs at runtime and predefining arguments in a file.
models/model.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f9aef3d7c3114f1c86695e119d64f7e83dcf6449716cb64df9d5b7b447adb8a5
3
  size 28419489
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:54066a13f090b94eb43c589c9090f9791810f74903998418c20e8502839d2c83
3
  size 28419489
models/model.ts CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:37d348addc879bc51d5e5fa89927ceb48c3521632135a0d5cdc488592961209e
3
- size 28620879
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10745ecb9c680c3bc2756da88326403153eee0964551a62da884d5fb21116f5e
3
+ size 28573133