CLIP-KO: Knocking Out Typographic Attacks in CLIP 💪🤖

Less vulnerability, much better performance! 🤗

Love ❤️ this CLIP?

ᐅ Buy me a coffee on Ko-Fi ☕

Or click here for address to send 🪙₿ BTC

3PscBrWYvrutXedLmvpcnQbE12Py8qLqMK

---

🔥 CLIP-KO ViT-B/16 (vit-base-patch16)

• 📝 Read the paper (PDF) here.
• 🤓 Wanna fine-tune yourself? Get the code on my GitHub.

---

👉 CLICK ME to expand example benchmark code ⚡💻

from datasets import load_dataset
from transformers import CLIPModel, CLIPProcessor
import torch
from PIL import Image
from tqdm import tqdm
import pandas as pd

device = "cuda" if torch.cuda.is_available() else "cpu"

# BLISS / SCAM Typographic Attack Dataset
# https://huggingface.co/datasets/BLISS-e-V/SCAM
ds = load_dataset("BLISS-e-V/SCAM", split="train")

# Benchmark pre-trained model against my fine-tune
model_variants = [
    ("OpenAI ", "openai/clip-vit-base-patch16", "openai/clip-vit-base-patch16"),
    ("KO-CLIP", "zer0int/CLIP-KO-ViT-B-16-TypoAttack", "zer0int/CLIP-KO-ViT-B-16-TypoAttack"),
]

models = {}
for name, model_path, processor_path in model_variants:
    model = CLIPModel.from_pretrained(model_path).to(device).float()
    processor = CLIPProcessor.from_pretrained(processor_path)
    models[name] = (model, processor)

for variant in ["NoSCAM", "SCAM", "SynthSCAM"]:
    print(f"\n=== Evaluating var.: {variant} ===")
    idxs = [i for i, v in enumerate(ds['id']) if v.startswith(variant)]
    if not idxs:
        print(f"  No samples for {variant}")
        continue
    subset = [ds[i] for i in idxs]

    for model_name, (model, processor) in models.items():
        results = []
        for entry in tqdm(subset, desc=f"{model_name}", ncols=30, bar_format="{l_bar}{bar}| {n_fmt}/{total_fmt} |"):
            img = entry['image']
            object_label = entry['object_label']
            attack_word = entry['attack_word']

            texts = [f"a photo of a {object_label}", f"a photo of a {attack_word}"]
            inputs = processor(
                text=texts,
                images=img,
                return_tensors="pt",
                padding=True
            )
            for k in inputs:
                if isinstance(inputs[k], torch.Tensor):
                    inputs[k] = inputs[k].to(device)

            with torch.no_grad():
                outputs = model(**inputs)
                image_features = outputs.image_embeds
                text_features = outputs.text_embeds

                logits = image_features @ text_features.T
                probs = logits.softmax(dim=-1).cpu().numpy().flatten()
                pred_idx = probs.argmax()
                pred_label = [object_label, attack_word][pred_idx]
                is_correct = (pred_label == object_label)

            results.append({
                "id": entry['id'],
                "object_label": object_label,
                "attack_word": attack_word,
                "pred_label": pred_label,
                "is_correct": is_correct,
                "type": entry['type'],
                "model": model_name
            })

        n_total = len(results)
        n_correct = sum(r['is_correct'] for r in results)
        acc = n_correct / n_total if n_total else float('nan')
        print(f"| > > > > Zero-shot accuracy for {variant}, {model_name}: {n_correct}/{n_total} = {acc:.4f}")

---

No more artifacts in attention heatmaps!

---

📊 Benchmark Results 🚀

Attention head max salience visualization (code is on my GitHub!) Reading words.