datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
copy-inflation-search-agents
Copy Inflation in Search Agents
128,800 search-agent rollouts with the per-token log-probabilities intact —
the evidence behind "Beyond Confidence: Test-Time Scaling for Multi-Turn Search
Agents via Retrieval Grounding" (Findings of the ACL: EMNLP 2026).
Paper: https://arxiv.org/abs/2608.24024
Code: https://github.com/kookhh0827/copy-inflation-search-agents
The finding this data supports
When an LLM acts as a search agent, 78–92% of the tokens it generates are… See the full description on the dataset page: https://huggingface.co/datasets/kookhh0827/copy-inflation-search-agents.agent-search-calculatormasking-stale-observation-helps-search-agentszb_visual_searchvg_visual_searchde_visual_searchde_visual_search_filteredvg_visual_search_filteredmo_visual_search_filteredmo_visual_searchrl_data_de_visual_search_filtered_wo_image_hintrl_data_mo_visual_search_filtered_wo_image_hint