datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
TopoSense-Bench
TopoSense-Bench: A Campus-Scale Benchmark for Semantic-Spatial Sensor Scheduling
TopoSense-Bench is a large-scale, rigorous benchmark designed to evaluate Large Language Models (LLMs) and agents on the Semantic-Spatial Sensor Scheduling (S³) problem. It features a realistic digital twin of a university campus equipped with 2,510 cameras and contains 5,250 natural language queries grounded in physical topology.
This dataset is the official benchmark for the ACM MobiCom 2026 paper:… See the full description on the dataset page: https://huggingface.co/datasets/IoT-Brain/TopoSense-Bench.brain-memory
🧠 NIFTY AI Agent: Memory OS Cloud Snapshot
Cloud backup repository for the NIFTY 50 Autonomous AI Agent Memory OS.
• Repository: nagarhimanshu37/brain-memory• Total Stored Records: 217• Last Synchronized: 2026-09-23 12:34:52 UTC
📊 Partition Statistics
Partition
Records
Description
conversation_memory
70
Multi-turn trader dialogues & intent logs
episodic_memory
50
Trading day episodes (facts vs interpretations)
experience_memory
50
Crystallized… See the full description on the dataset page: https://huggingface.co/datasets/nagarhimanshu37/brain-memory.code-training-il
Code-Training-IL
A 40,330-example instruction-tuning dataset for code: 20K Python (NVIDIA OpenCodeInstruct, test-filtered) + 20K TypeScript + 330 hand-written bilingual identity examples.
Overview
code-training-il is a curated, filtered instruction-tuning corpus for training small coding assistants. It is the dataset used to fine-tune code-il-E4B, a 4B on-device model.
The dataset was designed around a thesis: less data, better filtered, beats more data. The… See the full description on the dataset page: https://huggingface.co/datasets/BrainboxAI/code-training-il.medical-training-il
Medical-Training-IL
A bilingual (Hebrew / English) medical instruction-tuning corpus — curated for training small, on-device medical models for Israeli residents preparing for Stage A exams.
Overview
medical-training-il is a curated, bilingual medical instruction-tuning dataset designed to fine-tune language models for Israeli clinical reasoning. It combines high-quality English medical QA (USMLE-style, basic sciences, research-grounded) with ~5,000 Hebrew-native… See the full description on the dataset page: https://huggingface.co/datasets/BrainboxAI/medical-training-il.brainly
brainly.co.id dataset
Data Structure
The keys in each JSONL object include:
"id": An integer value representing the page of task from url (e.g. brainly.co.id/tugas/117).
"subject": A string indicating the subject of the question (e.g., "Fisika", "Matematika", "Sejarah").
"author": A string representing the author of the question.
"instruction": A string providing the instruction or prompt for the question.
"answerer_1", "answer_2": Strings representing the answerers for… See the full description on the dataset page: https://huggingface.co/datasets/lesserfield/brainly.uv-brain-s03_custom_with_rehearsalswebench-django
SWE-bench Django (30-task subset)
Thirty real bug-fix tasks from the Django project, derived from SWE-bench (Jimenez, Yang, et al., ICLR 2024). Each task is a merged pull request rewound to its buggy commit: an agent gets only the issue text, must locate and fix the bug in the codebase, and the fix is checked against the PR's held-out test.
This subset powers a Braintrust eval on behavior-vs-output scoring — whether a coding agent obeys a "locate code via vector search only"… See the full description on the dataset page: https://huggingface.co/datasets/BraintrustDataDev/swebench-django.uv-brain-s03_custom_with_rehearsal_v2brainstorming-ideation-sft-100k
Brainstorming and Ideation SFT (100K)
100,000 ShareGPT conversations demonstrating structured, high-quality brainstorming and ideation across 22 professional domains. Each example takes a realistic context and constraint, then generates specific, actionable, well-reasoned ideas — not generic advice dressed as creativity.
Motivation
Brainstorming and ideation is one of the highest-value use cases for AI assistants, and one where models routinely underperform:… See the full description on the dataset page: https://huggingface.co/datasets/stindardlogic/brainstorming-ideation-sft-100k.Pashto-Brain-Extraction-Dataset
🧠 Pashto Brain Extraction Dataset
A small experimental Pashto reasoning dataset designed to extract and preserve useful model reasoning/planning while discarding the final answer.
Keep the brain 🧠 — throw away the mouth 🗣️
🎯 Purpose
A language model may understand a question and produce useful reasoning while still generating poor, unnatural, or grammatically incorrect Pashto in its final answer.
Instead of throwing away the entire generation, this dataset… See the full description on the dataset page: https://huggingface.co/datasets/nassimjp/Pashto-Brain-Extraction-Dataset.ptv3-bericht-lora-de-300
ptv3-bericht-lora-de-300
Synthetic German dataset for fine-tuning LLMs to generate structured psychotherapy reports (PTV-3 / Bericht an den Gutachter) from therapy session transcripts.
Overview
Property
Value
Samples
311 (280 train / 31 val)
Language
German
Format
ChatML JSONL (system / user / assistant)
Teacher model
Qwen2.5-27B (local)
Generation
Two-stage: seed → session transcript → PTV-3 JSON report
Schema
Each sample… See the full description on the dataset page: https://huggingface.co/datasets/speed-brain-ai/ptv3-bericht-lora-de-300.brainblast-verified-footgun-corpus
Brainblast — Verified SDK Footgun Corpus (free sample)
The only code-training data that ships with a machine-checkable proof. Each
record is a real insecure→fixed code footgun with a replayable RED→GREEN
receipt: a deterministic checker fails the insecure version and passes the fixed
one. You don't trust the labels — you replay the proof.
This repo is a free 40-record sample (receipt-only tier). The full corpus is
4,183 proven records across 154 SDKs and 9 vulnerability classes… See the full description on the dataset page: https://huggingface.co/datasets/dsb117/brainblast-verified-footgun-corpus.LLM_BRAIn_datasetLLM_BRAIn: AI-driven Fast Generation of Robot Behaviour Tree based on Large Language Model
Original paper preprint: https://arxiv.org/abs/2305.19352
This paper introduces a pioneering methodology in autonomous robot control, denoted as LLM-BRAIn, enabling the generation of adaptive behaviors in robots in response to operator commands, while simultaneously considering a multitude of potential future events. LLM-BRAIn is a transformer-based Large Language Model (LLM) fine-tuned from the… See the full description on the dataset page: https://huggingface.co/datasets/ArtemLykov/LLM_BRAIn_dataset.xio-compliance-brain-triad-prompts
XIO Compliance Brain — Triad Reviewer Prompts
Reusable system prompts for running a multi-voice compliance debate against the same matter — the heart of XIO Compliance Brain's "Triad Review Engine" pattern.
This dataset extracts the production prompts from the open-source compliance-AI hackathon branch so others can replicate the Triad pattern (three reviewer voices + synthesis + optional Round 2) on any LLM that follows OpenAI-compatible chat APIs.
What's in this… See the full description on the dataset page: https://huggingface.co/datasets/slavazeph/xio-compliance-brain-triad-prompts.legal-training-il
Legal-Training-IL
A 17,613-example bilingual instruction-tuning corpus for Israeli legal reasoning — covering rulings, statutes, citizen-rights pages, and contract clauses.
Overview
legal-training-il is a curated, bilingual (Hebrew / English) instruction-tuning dataset designed to adapt general-purpose language models to Israeli legal work. It was built to train law-il-E2B, a 2B-parameter on-device legal assistant.
The dataset is not a scraped dump. Every example… See the full description on the dataset page: https://huggingface.co/datasets/BrainboxAI/legal-training-il.brainrot-conversation
Brainrot Conversation Dataset
The Brainrot Conversation dataset is a curated collection of synthetic multi turn dialogues designed to train conversational AI models in modern GenZ style expression. The dataset emphasizes chaotic humor, highly informal phrasing, and internet native slang commonly referred to as “brainrot”. It provides clean structure, consistent formatting, and sufficient volume to develop stylistically aligned chat models.
The dataset is suitable for fine tuning… See the full description on the dataset page: https://huggingface.co/datasets/grenishrai/brainrot-conversation.maindlock-brain-traces
mAIndlock — Brain-Region Deliberation Traces
Every NPC in mAIndlock is a
value-based decision network: six computational roles, each a real call to a small local
model (MiniCPM 1B for the sensing regions, Nemotron 3 Nano 4B for the voice), integrated by a
deterministic vmPFC. This dataset is the raw deliberation of those minds, recorded live
and fully offline.
Each row is one NPC turn and contains:
field
meaning
player_line
what the player said
regions[]
per region:… See the full description on the dataset page: https://huggingface.co/datasets/arbios/maindlock-brain-traces.brainrot-custom-dataset
Brainrot Fine-Tuning Dataset
This is a cleaned fine-tuning conversion of shvn22k/brainrot-dataset.
Target repo:
Nels2/brainrot-custom-dataset
Purpose
This dataset is intended for supervised fine-tuning experiments where the model learns to translate standard English into casual "brainrot" internet-speak style.
Format
Output format: chat
Each row uses a chat-style messages array:
{
"messages": [
{
"role": "system",
"content": "You… See the full description on the dataset page: https://huggingface.co/datasets/Nels2/brainrot-custom-dataset.hermes-brain-bench-v0.1
Hermes Brain Bench v0.1
A small, diverse, graded-rubric benchmark for picking the agent brain behind
a local-only Spark deployment of an OpenAI-compatible tool-using assistant
(developed against Hermes Agent, but
the suite is harness-agnostic — any OpenAI-tool-format runner works).
The bench answers a question that single-stream throughput benchmarks can't:
which local serving lane actually produces the more correct agent — under the
same rubric, run-to-run?
Companion artifacts… See the full description on the dataset page: https://huggingface.co/datasets/Orionfold/hermes-brain-bench-v0.1.Brain_Helix