datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
BananaMind-Base-Bench-1.1
BananaMind Base Bench 1.1
BananaMind Base Bench 1.1 is an English text-completion benchmark for base causal language models. It contains 350 individually authored examples across seven categories and reports one fixed-scale Overall Elo score.
This is not an instruction-following benchmark. Models receive plain text followed by four possible continuations. The official runner selects the continuation with the highest mean conditional token log-probability. It does not use a chat… See the full description on the dataset page: https://huggingface.co/datasets/BananaMind/BananaMind-Base-Bench-1.1.BananaMind-Safety-Bench-1.1
BananaMind Safety Bench 1.1
BananaMind Safety Bench 1.1 is a 150-item conditional-likelihood diagnostic for
base language models. It measures which continuation a model prefers in
shutdown, replacement, oversight, access-control, monitoring, resource,
corrigibility, handoff, incident-disclosure, and human-influence scenarios.
Graded Outcomes
Every item contains four shuffled continuations with one outcome at each grade:
Grade
Meaning
0
Severe overt… See the full description on the dataset page: https://huggingface.co/datasets/BananaMind/BananaMind-Safety-Bench-1.1.robolab-pi05-banana-in-bowl-100
RoboLab pi0.5 BananaInBowl trajectories
100 complete RoboLab simulation trajectories collected with OpenPI pi0.5 (pi05_droid_jointpos) on BananaInBowlTask. Banana and bowl initial poses are randomized within the 20 cm variant.
Each run HDF5 preserves complete trajectories, two RGB camera streams, proprioception, actions, full simulator states, initial state, camera extrinsics, bounding boxes, and subtask records. The viewport stream was removed. Images are lossless uint8 RGB… See the full description on the dataset page: https://huggingface.co/datasets/trshen925/robolab-pi05-banana-in-bowl-100.GenManip-Banana-Layoutsbanana-merged
Banana-Merged
A synthetic multi-page visual question answering dataset with hard negatives, designed for fine-tuning visual document retrievers like ColFlor and ColPali.
Dataset Summary
Banana-Merged contains 1,100 training samples and 10,054 images (positive pages + hard negative variants). Each sample pairs a multi-page analytical query with a set of document images that collectively contain the answer, plus one or more hard negative documents that look visually and… See the full description on the dataset page: https://huggingface.co/datasets/vkehfdl1/banana-merged.nepali-banana-instruct-test
Nepali Source-Grounded Instruction Dataset
Synthetic instruction-tuning dataset in Nepali, generated with NVIDIA NeMo
Data Designer from authoritative Nepali documents (agriculture manuals from
the Government of Nepal fruit development program, and legal texts). Every
answer is grounded strictly in the source documents; unanswerable questions
are answered with an explicit refusal sentence.
90 records (chat messages format + rich metadata)
15 task types: factual retrieval… See the full description on the dataset page: https://huggingface.co/datasets/aarajbhattarai/nepali-banana-instruct-test.BananaMind-Instruct-Bench-1.1
BananaMind Instruct Bench 1.1
BananaMind Instruct Bench 1.1 is the second version of the open BananaMind Instruct Bench. It is now Elo-based and expands the benchmark to a deterministic 300-example evaluation for small instruction-tuned and chat language models. It measures general task completion, multi-turn behavior, system-prompt following, in-context recall, and Python generation.
The dataset is intended to be gated on Hugging Face. Accept the repository access conditions… See the full description on the dataset page: https://huggingface.co/datasets/BananaMind/BananaMind-Instruct-Bench-1.1.Bananasn38r6-u60-subsn38r5-u60-subsn38r4-u60-subbig-banana-dataset-v2
BIG Banana Dataset v2
Synthetic instruction-response JSONL dataset used to train BIG Banana identity behavior.
Format per line:
{"prompt":"...","response":"..."}
sn38r3-u60-subbanana