--- license: mit library_name: open_clip pipeline_tag: zero-shot-image-classification base_model: openai/clip-vit-base-patch32 datasets: - juns/CS-CLIP-data tags: - clip - open_clip - vision-language - compositional-reasoning --- # CS-CLIP (ViT-B/32) CS-CLIP is CLIP ViT-B/32 fine-tuned on scene graph-guided hard negatives for robust compositional reasoning. It is from the paper [CS-CLIP: Compositional Scene Graph-guided CLIP for Robust Compositional Reasoning](https://arxiv.org/abs/2609.08242) (Findings of EMNLP 2026). - Code: https://github.com/statjuns2/CS-CLIP - Training data and benchmark files: [juns/CS-CLIP-data](https://huggingface.co/datasets/juns/CS-CLIP-data) The architecture is unchanged OpenAI CLIP ViT-B/32 (QuickGELU), stored in fp32. This repository holds both the OpenCLIP weights (`open_clip_model.safetensors`) and the Transformers weights (`model.safetensors`). ## Usage OpenCLIP: ```python import torch, open_clip from PIL import Image model, _, preprocess = open_clip.create_model_and_transforms("hf-hub:juns/CS-CLIP") tokenizer = open_clip.get_tokenizer("hf-hub:juns/CS-CLIP") model.eval() image = preprocess(Image.open("example.jpg")).unsqueeze(0) text = tokenizer(["a man riding a brown horse", "a brown man riding a horse"]) with torch.no_grad(): sims = model.encode_image(image, normalize=True) @ model.encode_text(text, normalize=True).T ``` Transformers: ```python from transformers import CLIPModel, CLIPProcessor model = CLIPModel.from_pretrained("juns/CS-CLIP") processor = CLIPProcessor.from_pretrained("juns/CS-CLIP") inputs = processor(text=["a man riding a brown horse", "a brown man riding a horse"], images=image, return_tensors="pt", padding=True) logits_per_image = model(**inputs).logits_per_image ``` ## Results Results of this checkpoint, reproduced with the evaluation scripts in the code repository: | SugarCrepe Rel / Obj / Att | Swap Obj / Att | Add Obj / Att | Avg | VG-Rel | VG-Att | COCO I2T / T2I R@1 | GQA two-choice | |---|---|---|---|---|---|---|---| | 83.71 / 94.31 / 87.94 | 78.05 / 81.98 | 90.83 / 89.02 | 88.59 | 85.96 | 77.36 | 53.20 / 42.17 | 71.27 | The paper's Table 1 reports the mean over three training seeds (SugarCrepe Avg 88.6 ± 0.2); this checkpoint is the seed-120 run. ## Training - **Initialization.** OpenAI CLIP ViT-B/32. - **Data.** 1,941,340 caption / hard-negative pairs on COCO train2014 images ([juns/CS-CLIP-data](https://huggingface.co/datasets/juns/CS-CLIP-data)). - **Objective.** Structure-CLIP's in-batch margin ranking loss (margin 0.2) plus 5 × a hard-negative hinge loss (margin 0.1). - **Optimization.** AdamW with a constant learning rate of 2e-6, weight decay 0.1, batch size 128 and 2 epochs, in fp32. ## Limitations Like other CLIP derivatives, the model may inherit biases from CLIP's web-scale pre-training data, and it was fine-tuned only on COCO. ## Citation ```bibtex @inproceedings{jeong2026csclip, title = {{CS-CLIP}: Compositional Scene Graph-guided {CLIP} for Robust Compositional Reasoning}, author = {Jeong, SeongJun and Jung, Minjoon and Choi, Woo Suk and Jang, Youwon and Zhang, Byoung-Tak}, booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2026}, year = {2026} } ```