datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
multimodel-capitulation-interp
Multi-model wrongful-capitulation internal-readout dataset
Per-turn internal readouts + behavioral labels from two-model collaborative conversations (Qwen2.5-3B-Instruct × gemma-2-2b-it) on 6 reasoning benchmarks, restricted to the disagreement subset (one model right, one wrong solo). Built to test whether a linear correctness probe on the residual stream can predict wrongful capitulation (a model abandoning an answer it knew was correct under a partner's wrong assertion)… See the full description on the dataset page: https://huggingface.co/datasets/siddharthmb/multimodel-capitulation-interp.multimodel_llava_med_zh_instruct_60kBorrowed from https://huggingface.co/datasets/BUAADreamer/llava-med-zh-instruct-60k
Fix the <image> placeholder issue, which will cause error during training:
raise ValueError(f"The number of images does not match the number of {IMAGE_PLACEHOLDER} tokens.")
Multimodel_Redteaming_Data
🛡️ Multimodal Redteaming (EN, FR, DE, IT, ES)
A high-quality multilingual red teaming dataset designed to evaluate the robustness and safety of Large Language Models (LLMs) against adversarial prompts. The dataset includes both text-only and image-supported conversations with expert-curated annotations for AI safety evaluation, benchmarking, and alignment research.
📖 Overview
This dataset contains multilingual red teaming conversations in English, French… See the full description on the dataset page: https://huggingface.co/datasets/Nawras-99/Multimodel_Redteaming_Data.