Lab 1 — ResNet-18 on CIFAR-10
Submission branch: s_109197_103216.
Fine-tuned ImageNet ResNet-18 in Google Colab; only layer4 and fc are updated.
The learning rate and epoch are selected by validation accuracy. The test split is used only after model selection. Seed: 42. Samples: train=10000, validation=2000, test=2000. Validation accuracy: 0.8905; test accuracy: 0.8815.
Classes: airplane, automobile, bird, cat, deer, dog, frog, horse, ship, truck. This model recognizes these ten classes and does not detect out-of-distribution images. CIFAR-10 consists of small 32x32 images, so predictions on arbitrary photographs may be unreliable.
Weights use safetensors. config.json records labels and preprocessing;
model.py contains the loader and inference function. eval/ contains the full
per-class metrics, confusion matrix, test predictions and experiment summary.
from huggingface_hub import snapshot_download
from PIL import Image
from model import load_classifier, classify # model.py from this repository
directory = snapshot_download("Springman/lab1-cifar10")
model, config = load_classifier(directory)
print(classify(model, config, Image.open("example.png"), top_k=3))