datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
k3
Kimi K3 Deduplicated Pi Agent Sessions
Native Hugging Face Agent Trace projection of greghavens/kimi-k3-coding-and-debugging-traces.
Unlike a row-level projection, this export first collapses the source dataset's
cumulative next-assistant prefixes. Each output .jsonl file represents one
complete source trajectory rather than one intermediate training prefix.
Build summary
Source revision: 33a874c3affbdb97e142752a9144e6624ef5bd07
Source cumulative rows: 3,956… See the full description on the dataset page: https://huggingface.co/datasets/C10X/k3.multiturn-chatomni-mathcn_k12SuperGPQADeepsearch-testspecialttttttwwwwwwwwtop5cepCodeFeedback-Filtered-Instruction-sharegptClaude-4.5-opus-high-reasoning-250x-under32kcladuefinewebteeeesc10x__Q-Pluse-details
Dataset Card for Evaluation run of c10x/Q-Pluse
Dataset automatically created during the evaluation run of model c10x/Q-Pluse
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An additional configuration… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/c10x__Q-Pluse-details.Gryphe-Claude_Sonnet-3.5-SlimOrca-140k-ShareGPT-Scoredtest1swrustdscladtestpdfv5longwriter-6k-filtered-sharegptifbalancedtest221000c10x__longthinker-details
Dataset Card for Evaluation run of c10x/longthinker
Dataset automatically created during the evaluation run of model c10x/longthinker
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An additional configuration… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/c10x__longthinker-details.kf