Update README.md

README.md

@@ -1,19 +1,64 @@
 ---
 license: mit
 tags:
-- automl
-- autogluon
-- image-classification
-- neural-network
+  - automl
+  - autogluon
+  - image-classification
+  - neural-network
+  - computer-vision
+  - education
 library_name: autogluon
 datasets:
-- ecopus/sign_identification   # replace if needed
+  - ecopus/sign_identification
+model-index:
+  - name: HW2 Neural AutoML — AutoGluon MultiModalPredictor (Signs)
+    results:
+      - task:
+          name: Image Classification
+          type: image-classification
+        dataset:
+          name: ecopus/sign_identification
+          type: ecopus/sign_identification
+          split: test
+        metrics:
+          - name: Accuracy
+            type: accuracy
+            value: 0.4286
+          - name: Macro-F1
+            type: f1
+            value: 0.3
 ---
 
-# HW2 Neural AutoML — AutoGluon MultiModalPredictor
+# HW2 Neural AutoML — AutoGluon MultiModalPredictor (Signs)
 
-Artifacts:
-- `ag_image_predictor.pkl` — predictor pickled with cloudpickle
-- `ag_image_predictor_dir.zip` — zipped native AutoGluon predictor directory
+**Course:** 24-679 — HW2: Models  
+**Task:** Supervised image classification on a classmate’s dataset (`ecopus/sign_identification`) using **AutoGluon Multimodal**.  
+**Goal:** Build a compact model under a small compute budget and report results with a clear, reproducible pipeline.
+
+
+## Summary
+
+- **Backbone:** `resnet18` (via timm)
+- **Input resolution:** 224×224 (images resized in Colab)
+- **Train/Val/Test:** ~64% / 16% / 20% split (stratified)
+- **Epochs:** 3 (short budget, early-stop not overridden)
+- **Batch size:** 8
+- **Metric (val):** Accuracy + Macro-F1
+- **Result (test):** Accuracy = **0.4286**, Macro-F1 = **0.3**
+
+## Dataset
+
+- **Source:** `ecopus/sign_identification`
+- **Task:** Multiclass sign recognition
+- **Classes:** [N_CLASSES] (e.g., list them if short)
+- **Preprocessing:** 
+  - `datasets` → decode to `PIL`
+  - Resize to 224×224, RGB
+  - Labels normalized to integers/strings for AutoGluon
+
+## Training & AutoML Setup
+
+**Library:** `autogluon.multimodal.MultiModalPredictor`  
+**Problem type:** `multiclass`  
+**Eval metric:** `accuracy` (Macro-F1 also reported)  
 
-Trained for HW2 (image classification) using a classmate's dataset.