Datasets:

TreezzZ
/

SMILE-data

pretty_name: SMILE artifacts
language:
  - en
tags:
  - in-context-learning
  - vision-language-models
  - hard-negatives
task_categories:
  - visual-question-answering
  - image-classification
viewer: false

SMILE: CLIP embeddings, mined hard-negative pools and predictions

Artifacts for The Value of Being Wrong: Self-Mined Visual In-Context Learning from Errors (NeurIPS 2026 Spotlight). Paper: arXiv:2610.16977. Code: github.com/Tree-Shu-Zhao/SMILE.

clip_embeddings/
  <data>_<split>_image.pt            # CLIP ViT-g-14 (laion2b_s12b_b42k) image embeddings, L2-normalized
  <data>_<split>_ids.pt              # the sample id of each row
hard_negatives/<data>/<org_model>/
  hard_negatives.jsonl               # the pool: training samples the model gets wrong, with its voted prediction
  hard_negatives_predictions.jsonl   # the voted prediction for every training sample
  hard_negatives_stats.json
  hard_negatives_params.json
predictions/<data>/<org_model>/
  vicl_<k>_<negative_k>.jsonl        # test predictions, with retrieved_ids and contrastive_ids
  vicl_<k>_<negative_k>_params.json

<data>: cub, flowers, okvqa, vizwiz. <split>: train, test (for VizWiz, test is its val split).
<org_model>: Qwen_Qwen2.5-VL-7B-Instruct, Qwen_Qwen2.5-VL-3B-Instruct, google_gemma-3-12b-it, google_gemma-3-4b-it.
Pool records: id, question, imgpath, gt_answer, voted_pred, vote_count, total_votes, token_f1, is_correct.
Prediction records: id, question, imgpath, pred_answer, gt_answer, k, retrieved_ids, contrastive_ids. The VizWiz records hold id, pred_answer, query_prediction, retrieved_ids and contrastive_ids.
The predictions are one inference run per model and dataset. VizWiz predictions are stored before the answerability vote, which the scorer applies.
imgpath is relative to the root of the code repository (data/<dataset>/...).
The records contain questions and answers from CUB-200-2011, Oxford 102 Flowers, OK-VQA and VizWiz, which remain under the terms of those datasets. No images are included.

Usage

From the root of the code repository:

uvx --from "huggingface_hub>=1.0" hf download TreezzZ/SMILE-data --repo-type dataset --local-dir . \
  --include "clip_embeddings/*" --include "hard_negatives/*" --include "predictions/*"

The files land where the code reads them. To score the released predictions:

bash scripts/cub/evaluate.sh predictions/cub/Qwen_Qwen2.5-VL-7B-Instruct/vicl_8_4.jsonl
bash scripts/vizwiz/evaluate.sh predictions/vizwiz/Qwen_Qwen2.5-VL-7B-Instruct/vicl_12_8.jsonl

Citation

@misc{zhao2026smile,
  title={The Value of Being Wrong: Self-Mined Visual In-Context Learning from Errors},
  author={Shu Zhao and Tan Yu},
  year={2026},
  eprint={2610.16977},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2610.16977},
}

SMILE: CLIP embeddings, mined hard-negative pools and predictions

Contents

Usage

Citation