Upload DeiT model from experiment a1

.gitattributes

@@ -33,3 +33,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+training_curves/DeiT3_combined_metrics.png filter=lfs diff=lfs merge=lfs -text
+training_curves/DeiT_combined_metrics.png filter=lfs diff=lfs merge=lfs -text
+training_notebook_a1.ipynb filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,161 @@
+---
+license: apache-2.0
+tags:
+- vision-transformer
+- image-classification
+- pytorch
+- timm
+- deit
+- gravitational-lensing
+- strong-lensing
+- astronomy
+- astrophysics
+datasets:
+- C21
+metrics:
+- accuracy
+- auc
+- f1
+model-index:
+- name: DeiT-a1
+  results:
+  - task:
+      type: image-classification
+      name: Strong Gravitational Lens Discovery
+    dataset:
+      type: common-test-sample
+      name: Common Test Sample (More et al. 2024)
+    metrics:
+    - type: accuracy
+      value: 0.8007
+      name: Average Accuracy
+    - type: auc
+      value: 0.8275
+      name: Average AUC-ROC
+    - type: f1
+      value: 0.5055
+      name: Average F1-Score
+---
+
+# 🌌 deit-gravit-a1
+
+🔭 This model is part of **GraViT**: Transfer Learning with Vision Transformers and MLP-Mixer for Strong Gravitational Lens Discovery
+
+🔗 **GitHub Repository**: [https://github.com/parlange/gravit](https://github.com/parlange/gravit)
+
+## 🛰️ Model Details
+
+- **🤖 Model Type**: DeiT
+- **🧪 Experiment**: A1 - C21-classification-head
+- **🌌 Dataset**: C21
+- **🪐 Fine-tuning Strategy**: classification-head
+
+
+
+## 💻 Quick Start
+
+```python
+import torch
+import timm
+
+# Load the model directly from the Hub
+model = timm.create_model(
+    'hf-hub:parlange/deit-gravit-a1',
+    pretrained=True
+)
+model.eval()
+
+# Example inference
+dummy_input = torch.randn(1, 3, 224, 224)
+with torch.no_grad():
+    output = model(dummy_input)
+    predictions = torch.softmax(output, dim=1)
+print(f"Lens probability: {predictions[0][1]:.4f}")
+```
+
+## ⚡️ Training Configuration
+
+**Training Dataset:** C21 (Cañameras et al. 2021)  
+**Fine-tuning Strategy:** classification-head
+
+
+| 🔧 Parameter | 📝 Value |
+|--------------|----------|
+| Batch Size | 192 |
+| Learning Rate | AdamW with ReduceLROnPlateau |
+| Epochs | 100 |
+| Patience | 10 |
+| Optimizer | AdamW |
+| Scheduler | ReduceLROnPlateau |
+| Image Size | 224x224 |
+| Fine Tune Mode | classification_head |
+| Stochastic Depth Probability | 0.1 |
+
+
+## 📈 Training Curves
+
+![Combined Training Metrics](https://huggingface.co/parlange/deit-gravit-a1/resolve/main/training_curves/DeiT_combined_metrics.png)
+
+
+## 🏁 Final Epoch Training Metrics
+
+| Metric | Training | Validation |
+|:---------:|:-----------:|:-------------:|
+| 📉 Loss | 0.1886 | 0.2568 |
+| 🎯 Accuracy | 0.9267 | 0.9130 |
+| 📊 AUC-ROC | 0.9788 | 0.9631 |
+| ⚖️ F1 Score | 0.9268 | 0.9136 |
+
+
+## ☑️ Evaluation Results
+
+### ROC Curves and Confusion Matrices
+
+Performance across all test datasets (a through l) in the Common Test Sample (More et al. 2024):
+
+![ROC + Confusion Matrix - Dataset A](https://huggingface.co/parlange/deit-gravit-a1/resolve/main/roc_confusion_matrix/DeiT_roc_confusion_matrix_a.png)
+![ROC + Confusion Matrix - Dataset B](https://huggingface.co/parlange/deit-gravit-a1/resolve/main/roc_confusion_matrix/DeiT_roc_confusion_matrix_b.png)
+![ROC + Confusion Matrix - Dataset C](https://huggingface.co/parlange/deit-gravit-a1/resolve/main/roc_confusion_matrix/DeiT_roc_confusion_matrix_c.png)
+![ROC + Confusion Matrix - Dataset D](https://huggingface.co/parlange/deit-gravit-a1/resolve/main/roc_confusion_matrix/DeiT_roc_confusion_matrix_d.png)
+![ROC + Confusion Matrix - Dataset E](https://huggingface.co/parlange/deit-gravit-a1/resolve/main/roc_confusion_matrix/DeiT_roc_confusion_matrix_e.png)
+![ROC + Confusion Matrix - Dataset F](https://huggingface.co/parlange/deit-gravit-a1/resolve/main/roc_confusion_matrix/DeiT_roc_confusion_matrix_f.png)
+![ROC + Confusion Matrix - Dataset G](https://huggingface.co/parlange/deit-gravit-a1/resolve/main/roc_confusion_matrix/DeiT_roc_confusion_matrix_g.png)
+![ROC + Confusion Matrix - Dataset H](https://huggingface.co/parlange/deit-gravit-a1/resolve/main/roc_confusion_matrix/DeiT_roc_confusion_matrix_h.png)
+![ROC + Confusion Matrix - Dataset I](https://huggingface.co/parlange/deit-gravit-a1/resolve/main/roc_confusion_matrix/DeiT_roc_confusion_matrix_i.png)
+![ROC + Confusion Matrix - Dataset J](https://huggingface.co/parlange/deit-gravit-a1/resolve/main/roc_confusion_matrix/DeiT_roc_confusion_matrix_j.png)
+![ROC + Confusion Matrix - Dataset K](https://huggingface.co/parlange/deit-gravit-a1/resolve/main/roc_confusion_matrix/DeiT_roc_confusion_matrix_k.png)
+![ROC + Confusion Matrix - Dataset L](https://huggingface.co/parlange/deit-gravit-a1/resolve/main/roc_confusion_matrix/DeiT_roc_confusion_matrix_l.png)
+
+### 📋 Performance Summary
+
+Average performance across 12 test datasets from the Common Test Sample (More et al. 2024):
+
+| Metric | Value |
+|-----------|----------|
+| 🎯 Average Accuracy | 0.8007 |
+| 📈 Average AUC-ROC | 0.8275 |
+| ⚖️ Average F1-Score | 0.5055 |
+
+
+## 📘 Citation
+
+If you use this model in your research, please cite:
+
+```bibtex
+@misc{parlange2025gravit,
+      title={GraViT: Transfer Learning with Vision Transformers and MLP-Mixer for Strong Gravitational Lens Discovery}, 
+      author={René Parlange and Juan C. Cuevas-Tello and Octavio Valenzuela and Omar de J. Cabrera-Rosas and Tomás Verdugo and Anupreeta More and Anton T. Jaelani},
+      year={2025},
+      eprint={2509.00226},
+      archivePrefix={arXiv},
+      primaryClass={cs.CV},
+      url={https://arxiv.org/abs/2509.00226}, 
+}
+```
+
+---
+
+
+## Model Card Contact
+
+For questions about this model, please contact the author through: https://github.com/parlange/

config.json

@@ -0,0 +1,76 @@
+{
+  "architecture": "vit_base_patch16_224",
+  "num_classes": 2,
+  "num_features": 1000,
+  "global_pool": "avg",
+  "crop_pct": 0.875,
+  "interpolation": "bicubic",
+  "mean": [
+    0.485,
+    0.456,
+    0.406
+  ],
+  "std": [
+    0.229,
+    0.224,
+    0.225
+  ],
+  "first_conv": "conv1",
+  "classifier": "fc",
+  "input_size": [
+    3,
+    224,
+    224
+  ],
+  "pool_size": [
+    7,
+    7
+  ],
+  "pretrained_cfg": {
+    "tag": "gravit_a1",
+    "custom_load": false,
+    "input_size": [
+      3,
+      224,
+      224
+    ],
+    "fixed_input_size": true,
+    "interpolation": "bicubic",
+    "crop_pct": 0.875,
+    "crop_mode": "center",
+    "mean": [
+      0.485,
+      0.456,
+      0.406
+    ],
+    "std": [
+      0.229,
+      0.224,
+      0.225
+    ],
+    "num_classes": 2,
+    "pool_size": [
+      7,
+      7
+    ],
+    "first_conv": "conv1",
+    "classifier": "fc"
+  },
+  "model_name": "deit_gravit_a1",
+  "experiment": "a1",
+  "training_strategy": "classification-head",
+  "dataset": "C21",
+  "hyperparameters": {
+    "batch_size": "192",
+    "learning_rate": "AdamW with ReduceLROnPlateau",
+    "epochs": "100",
+    "patience": "10",
+    "optimizer": "AdamW",
+    "scheduler": "ReduceLROnPlateau",
+    "image_size": "224x224",
+    "fine_tune_mode": "classification_head",
+    "stochastic_depth_probability": "0.1"
+  },
+  "hf_hub_id": "parlange/deit-gravit-a1",
+  "license": "apache-2.0"
+}

deit-gravit-a1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24a921a917a65d38c27e1fed1091ad14c48152cfcbaf0b9bf502f69691061418
+size 343259194

evaluation_results.csv

@@ -0,0 +1,133 @@
+Model,Dataset,Loss,Accuracy,AUCROC,F1
+ViT,a,0.2907674585365717,0.8663942156554543,0.8790764272559853,0.36281859070464767
+ViT,b,0.3363787103321372,0.8409305249921408,0.8598268876611419,0.3235294117647059
+ViT,c,0.3624865817540249,0.8145237346746306,0.8460092081031308,0.29086538461538464
+ViT,d,0.20147570416317237,0.9100911662999057,0.914537753222836,0.4583333333333333
+ViT,e,0.38351454359247183,0.8397365532381997,0.8551653674411563,0.6237113402061856
+ViT,f,0.28049477608445245,0.8673224382309659,0.873733035896412,0.12378516624040921
+ViT,g,0.20559095215797424,0.9115,0.982819,0.9165225593460148
+ViT,h,0.21943247628211976,0.8975,0.9817622222222222,0.9045771916214119
+ViT,i,0.13406987351179123,0.9481666666666667,0.9929653333333334,0.9493567822830158
+ViT,j,1.5079676084518432,0.4855,0.5106800555555556,0.18869908015768724
+ViT,k,1.4364465267062188,0.5221666666666667,0.6366698888888889,0.200278940027894
+ViT,l,0.6356719980429736,0.7652688911215695,0.7653929261211886,0.6046842995814409
+MLP-Mixer,a,0.23711907013393504,0.904118201823326,0.8973158379373849,0.40545808966861596
+MLP-Mixer,b,0.32535140213071456,0.8626218170386671,0.8485451197053407,0.32248062015503876
+MLP-Mixer,c,0.26777226502923085,0.8937441056271612,0.8793204419889503,0.38095238095238093
+MLP-Mixer,d,0.24414238826503773,0.9034894687205282,0.8925340699815838,0.40388349514563104
+MLP-Mixer,e,0.4684625132219982,0.7826564215148188,0.7941497010519943,0.5123152709359606
+MLP-Mixer,f,0.25652938471803655,0.9010920920145612,0.8745385460021788,0.14006734006734006
+MLP-Mixer,g,0.21773884654045106,0.9175,0.981478,0.9204819277108434
+MLP-Mixer,h,0.18721229648590088,0.934,0.987164111111111,0.9353574926542605
+MLP-Mixer,i,0.17468452334403992,0.9391666666666667,0.9881612222222221,0.9401148482362592
+MLP-Mixer,j,1.2574891588687898,0.499,0.5052343333333333,0.19063004846526657
+MLP-Mixer,k,1.2144348313808442,0.5206666666666667,0.6023342222222222,0.19754464285714285
+MLP-Mixer,l,0.5478135314681644,0.7854158955105495,0.7718881621999252,0.6208893871449925
+CvT,a,0.48862357091431496,0.7661112857591952,0.7645202578268876,0.23613963039014374
+CvT,b,0.606147515758783,0.6821754165356806,0.7114585635359115,0.185334407735697
+CvT,c,0.5599785684152655,0.7161270040867652,0.7296648250460406,0.2030008826125331
+CvT,d,0.2567151319639595,0.8965734045897517,0.8801289134438306,0.41144901610017887
+CvT,e,0.634658475738718,0.677277716794731,0.7027472943313403,0.4389312977099237
+CvT,f,0.4719573438800043,0.7663232902176439,0.767503786678703,0.07083461656914075
+CvT,g,0.37052693724632263,0.824,0.9481419444444444,0.845478489903424
+CvT,h,0.34604967188835145,0.842,0.9550463333333333,0.8590544157002676
+CvT,i,0.18526951444149017,0.9376666666666666,0.9862493333333333,0.9392067620286085
+CvT,j,1.7297569365501404,0.37716666666666665,0.23870022222222223,0.10016855285335902
+CvT,k,1.5444995085000992,0.49083333333333334,0.47829777777777777,0.1198501872659176
+CvT,l,0.7996079008316138,0.686954682459944,0.637911609367734,0.5204147764095917
+Swin,a,0.41406949968579354,0.8063502043382584,0.9315,0.346072186836518
+Swin,b,0.46100307634250803,0.7742848160955674,0.9219889502762431,0.31226053639846746
+Swin,c,0.4275265830337383,0.7950330084878969,0.9307624309392264,0.3333333333333333
+Swin,d,0.12031445050063429,0.9597610814209369,0.9857863720073665,0.7180616740088106
+Swin,e,0.7676329288210487,0.6114160263446762,0.8420797699235602,0.47941176470588237
+Swin,f,0.39017997524092174,0.8144218108589575,0.9367503135673768,0.11976487876561352
+Swin,g,0.2684375742673874,0.8763333333333333,0.9830383333333332,0.8885551216581556
+Swin,h,0.25068944895267486,0.8873333333333333,0.9861691111111113,0.8974514563106796
+Swin,i,0.08781581997871399,0.9746666666666667,0.9978362222222223,0.974950560316414
+Swin,j,2.0428302302360533,0.42033333333333334,0.3025185555555555,0.11320754716981132
+Swin,k,1.8622084745168686,0.5186666666666667,0.5179311666666666,0.1332533013205282
+Swin,l,0.8495750360742048,0.7241816931944371,0.6664912098538803,0.5617543270038649
+CaiT,a,0.40013275007637966,0.8390443256837472,0.9108057090239411,0.3694581280788177
+CaiT,b,0.36606653323716165,0.8506758880855076,0.9150920810313077,0.3870967741935484
+CaiT,c,0.45299853780328836,0.8003772398616787,0.8942338858195211,0.32085561497326204
+CaiT,d,0.1995404364271233,0.9261238604212512,0.9594475138121547,0.5607476635514018
+CaiT,e,0.4928394595302159,0.7771679473106476,0.8707636418678574,0.5964214711729622
+CaiT,f,0.36103595997415405,0.850050344667338,0.9170773784465286,0.13416815742397137
+CaiT,g,0.2526822371482849,0.9083333333333333,0.976861222222222,0.9132218365414957
+CaiT,h,0.2987706809043884,0.8816666666666667,0.9732373333333334,0.8907356109572176
+CaiT,i,0.16439563977718352,0.9483333333333334,0.9884212222222222,0.9491636602164644
+CaiT,j,1.6938938024044037,0.48483333333333334,0.34579655555555555,0.18593626547274164
+CaiT,k,1.6056072112321853,0.5248333333333334,0.5246175555555556,0.1984818667416362
+CaiT,l,0.7488868967172212,0.7520490719686954,0.6676479234122561,0.5916572324305495
+DeiT,a,0.4765459399334255,0.7749135491983653,0.850939226519337,0.28112449799196787
+DeiT,b,0.29641423940058964,0.8814838101226029,0.9148747697974217,0.426179604261796
+DeiT,c,0.4700488794977955,0.7830870795347376,0.8519852670349909,0.28865979381443296
+DeiT,d,0.11101193318283434,0.9657340458975165,0.9766685082872928,0.7197943444730077
+DeiT,e,0.4436056611305011,0.8035126234906695,0.8648906380080225,0.6100217864923747
+DeiT,f,0.337409115804358,0.8522964913639532,0.8966829128564795,0.12802926383173296
+DeiT,g,0.1696254106760025,0.9383333333333334,0.9919374444444445,0.9412884798476674
+DeiT,h,0.26168069899082186,0.8861666666666667,0.9872772222222223,0.8967498110355253
+DeiT,i,0.07133128488063813,0.983,0.998361111111111,0.9830957905203845
+DeiT,j,1.940858449459076,0.4736666666666667,0.36180238888888894,0.10130904951622083
+DeiT,k,1.8425643298625947,0.5183333333333333,0.5667426111111111,0.10967344423906346
+DeiT,l,0.8096289309307544,0.7481360054994448,0.6678239118866791,0.5796487512134851
+DeiT3,a,0.3153771912384543,0.8786545111600126,0.8877375690607736,0.396875
+DeiT3,b,0.27676928310377796,0.8972021376925495,0.9046813996316758,0.43717728055077454
+DeiT3,c,0.3605959014448869,0.8566488525620874,0.867316758747698,0.35774647887323946
+DeiT3,d,0.23533754803987467,0.9091480666457089,0.9213038674033148,0.4677716390423573
+DeiT3,e,0.42457847907305024,0.8111964873765093,0.853205176719897,0.596244131455399
+DeiT3,f,0.28918993525220316,0.8904809852064132,0.8928482767899381,0.15227817745803357
+DeiT3,g,0.17661127078533173,0.9428333333333333,0.9877949999999999,0.9447041753990005
+DeiT3,h,0.2210533607006073,0.9213333333333333,0.9848492222222223,0.9254579911560329
+DeiT3,i,0.1546455579996109,0.9491666666666667,0.9909291111111113,0.9505271695052717
+DeiT3,j,1.796478444457054,0.5065,0.4410062222222223,0.17405857740585773
+DeiT3,k,1.774512730240822,0.5128333333333334,0.45727222222222225,0.17592331547786863
+DeiT3,l,0.7417711890545635,0.7793876579768388,0.6962303135333643,0.6175985334555454
+Twins_SVT,a,0.4130411987707923,0.8233259981138007,0.891621546961326,0.3403755868544601
+Twins_SVT,b,0.3827307312247214,0.8201823325998113,0.8948416206261509,0.33642691415313225
+Twins_SVT,c,0.4311113485054039,0.8157812008802263,0.8846832412523021,0.3310502283105023
+Twins_SVT,d,0.07931527397245478,0.9814523734674631,0.9895819521178638,0.830945558739255
+Twins_SVT,e,0.5826191509594378,0.7189901207464325,0.8197532732914554,0.5311355311355311
+Twins_SVT,f,0.3374182524886806,0.8535357447138099,0.9097097385998186,0.13296652911508483
+Twins_SVT,g,0.22945928347110747,0.904,0.9866478888888889,0.9113300492610837
+Twins_SVT,h,0.25510910618305205,0.9016666666666666,0.9884432222222223,0.9093701996927803
+Twins_SVT,i,0.06859854513406753,0.9895,0.9990906666666667,0.9894701654688284
+Twins_SVT,j,2.877535055875778,0.41883333333333334,0.1655298888888889,0.027336122733612273
+Twins_SVT,k,2.7166742979884146,0.5043333333333333,0.4133546666666667,0.031901041666666664
+Twins_SVT,l,1.0960884158819697,0.7418433715826768,0.5939750429978701,0.5637176050044683
+Twins_PCPVT,a,0.44722607902332284,0.7843445457403332,0.8699972375690609,0.2883817427385892
+Twins_PCPVT,b,0.34229608893356994,0.854133920150896,0.9080570902394106,0.3746630727762803
+Twins_PCPVT,c,0.47921196653085174,0.7686262181703867,0.8568545119705341,0.27416173570019725
+Twins_PCPVT,d,0.2515370008031815,0.913234831813895,0.9428913443830571,0.5018050541516246
+Twins_PCPVT,e,0.48251375945287006,0.7661909989023051,0.858896541285098,0.5661914460285132
+Twins_PCPVT,f,0.3832702880289431,0.8290604910541399,0.8924112354771649,0.11187122736418512
+Twins_PCPVT,g,0.23884364533424376,0.9121666666666667,0.9764792222222223,0.9165743232547096
+Twins_PCPVT,h,0.31143191051483154,0.8668333333333333,0.9660443333333333,0.8787372894217635
+Twins_PCPVT,i,0.1907261505126953,0.9435,0.9881127777777777,0.9446891825746452
+Twins_PCPVT,j,1.5205544946193694,0.48783333333333334,0.4457817777777778,0.1850967913020419
+Twins_PCPVT,k,1.472436999797821,0.5191666666666667,0.4698568888888889,0.19480881942506278
+Twins_PCPVT,l,0.7133773285155887,0.7375601501771456,0.6902435027241198,0.5768607724443686
+PiT,a,0.4122267626378192,0.8280414963847846,0.8927302025782688,0.34803337306317045
+PiT,b,0.39009387647344873,0.8230116315624018,0.8981012891344383,0.34152046783625734
+PiT,c,0.4812444057415132,0.7843445457403332,0.8707458563535911,0.2985685071574642
+PiT,d,0.08882246825332275,0.9710782772712984,0.9849594843462247,0.7604166666666666
+PiT,e,0.498657209813006,0.7705817782656421,0.8615000378415197,0.5828343313373253
+PiT,f,0.3495647843132805,0.848423824645651,0.9087592713952772,0.1298354824366385
+PiT,g,0.21983683621883393,0.9066666666666666,0.9891565555555555,0.9137931034482759
+PiT,h,0.2681618107557297,0.8861666666666667,0.988594888888889,0.8968122072820668
+PiT,i,0.06011278629302978,0.9851666666666666,0.9989021111111112,0.9852282157676349
+PiT,j,2.38970015335083,0.4266666666666667,0.21387577777777778,0.048672566371681415
+PiT,k,2.2299761089086534,0.5051666666666667,0.5063448888888888,0.055961844197138316
+PiT,l,0.9434709910593522,0.7408386653270583,0.6225121253803022,0.5664750110570544
+Ensemble,a,,0.9107198994027036,0.9407062615101288,0.5086505190311419
+Ensemble,b,,0.8880855077019805,0.9347661141804788,0.4523076923076923
+Ensemble,c,,0.8792832442628105,0.928644567219153,0.4336283185840708
+Ensemble,d,,0.9745363093366866,0.9857882136279925,0.784
+Ensemble,e,,0.8386388583973655,0.901748278210853,0.6666666666666666
+Ensemble,f,,0.9135620788474944,0.9448540229934943,0.20851063829787234
+Ensemble,g,,0.9431666666666667,0.9957611111111111,0.9458987783595113
+Ensemble,h,,0.9385,0.9963097777777778,0.9417153688200917
+Ensemble,i,,0.989,0.9993681111111111,0.9890510948905109
+Ensemble,j,,0.466,0.33056166666666664,0.0686046511627907
+Ensemble,k,,0.5118333333333334,0.5398214444444445,0.074565560821485
+Ensemble,l,,0.7875839458516207,0.6703077926895804,0.6177562089637454

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:153232e680b4264d1851df05ec05af158e61fda6d3a74e9fd902d8d027c3e7f0
+size 343214864

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24a921a917a65d38c27e1fed1091ad14c48152cfcbaf0b9bf502f69691061418
+size 343259194