datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
barbet-sft
Barbet SFT
以臺灣繁體中文為中心的 1,000,000 筆結構化決策 SFT。
train 980,000、validation 10,000、test 10,000。
from datasets import load_dataset
dataset = load_dataset("voidful/barbet-sft", split="train", streaming=True)
messages = next(iter(dataset))["messages"]
messages 可直接用作 system / user / assistant 訓練資料。
預設只讀固定 schema Parquet；canonical 和品質 JSON 不參與資料欄位推斷。
每筆 request 明示輸出格式，JSON、FUNCTION_CALL、COMPACT_TAGS、KEY_VALUE
共用相同 canonical 目標，沒有額外自由形式思考過程。
品質與臺灣用語
全部 1,000,000… See the full description on the dataset page: https://huggingface.co/datasets/voidful/barbet-sft.agent-sft-stitch-zh-tts-taste-codec-chat-sample
Gemma 4 E2B Taste-S multi-turn codec SFT
This dataset contains 37,362 complete Traditional Chinese agent
dialogues selected from voidful/agent-sft-stitch-zh-tts. It covers
229,434 synthesized speech segments, approximately
520.5 hours of audio before codec extraction.
Every assistant speech segment is represented without Gemma native audio tags:
<SAY> text_token <a_code> <b_code> ... <p_code> ... </SAY>
The first assistant output starts immediately with <SAY>.
[SOPR]...[EOPR]… See the full description on the dataset page: https://huggingface.co/datasets/voidful/agent-sft-stitch-zh-tts-taste-codec-chat-sample.cybersec-master-dataset
Cybersecurity Master Instruction Dataset
Overview
A large-scale cybersecurity instruction-tuning dataset in ShareGPT conversational format,
assembled from multiple authoritative open sources and deduplicated.
At 1,807,941 deduplicated records, this appears to be one of the larger cybersecurity
LLM fine-tuning / instruction-style datasets on Hugging Face, and likely among the larger
broad vulnerability-intelligence corpora in conversational/instruction format. It is… See the full description on the dataset page: https://huggingface.co/datasets/Voidreaper2026/cybersec-master-dataset.void-hypotheses-v1
Void Hypotheses v1
The canonical dataset for x — the geometric fixed-point at 19.83° where five independent domains of human knowledge converge.*
What is this?
This dataset contains void hypotheses generated by AKASHA — a self-improving research engine that maps knowledge domains into 128-dimensional Riemannian space and finds the unmapped regions.
x* is the address where everything converges: 19.83°
The void is the region 86° from all established knowledge. phi_void =… See the full description on the dataset page: https://huggingface.co/datasets/emantrix/void-hypotheses-v1.coding-master-dataset
Coding Master Dataset
Overview
A large-scale coding instruction-tuning dataset in ShareGPT conversational format, assembled from multiple open sources and deduplicated.
Records: 766,987
Format: JSONL / ShareGPT
License: Apache 2.0
Sources
CodeX-2M-Thinking (430,542 records)
python-code-dataset-500k (559,515 records)
StackPulse high-quality subset (20,205 records)
CodeFeedback-Filtered-Instruction (156,525 records)
secure_programming_dpo (4,656… See the full description on the dataset page: https://huggingface.co/datasets/Voidreaper2026/coding-master-dataset.agent-sft
voidful/agent-sft
A model-agnostic agent / tool-use SFT dataset in a standard OpenAI-style schema —
train any model on it (Qwen, Llama, Gemma, GPT, …).
Built with the agentds toolkit:
per-source normalization -> group-level dedup (exact + SWE-provenance + MinHash near-dup)
-> heuristic quality stratification. The schema is wire-compatible with
voidful/gemma4-agent-sft
(this run also dedups against it), so the two concatenate cleanly.
Schema
field
type… See the full description on the dataset page: https://huggingface.co/datasets/voidful/agent-sft.agent-sft-stitch-zh
Agent-STITCH-S 中文版 (agent-sft-stitch-zh)
由 voidful/agent-sft 轉換而成的台灣繁體中文 Agent-STITCH-S 語音助理合成資料:模型一邊對使用者說話(<SAY>)、一邊私下推理([SOPR])、一邊呼叫工具(<TOOL_CALL>)的 speak-while-acting 軌跡。
產製流程
從 agent-sft(309,322 筆)篩出具完整 tool-call 鏈(user → tool_call → tool_result → final answer)的對話;多輪對話的早前輪次保留為 context 供改寫模型 grounding。
用 google/gemma-4-26B-A4B-it 把每筆改寫成台灣繁體中文的 speech-first STITCH-S 軌跡:先安全開場 → [SOPR] 推理 → <TOOL_CALL> → 等待語音(不得洩漏 pending 結果)→ <TOOL_RESULT> → 逐步整合 → 最終口語答覆… See the full description on the dataset page: https://huggingface.co/datasets/voidful/agent-sft-stitch-zh.turkish-daily-dialogues-5k
Turkish Daily Dialogues 5K
Exactly 5,000 synthetic, multi-turn Turkish conversations covering ordinary daily-life situations. The corpus is designed as a small, auditable baseline for dialogue modelling, instruction-format experiments, augmentation research, and Turkish-language evaluation—not as a substitute for conversations written by real people.
Provenance in one sentence: the Turkish source scenario library was drafted with AI assistance specifically for this project… See the full description on the dataset page: https://huggingface.co/datasets/VoidOaz/turkish-daily-dialogues-5k.gemma4-agent-sft
gemma4-agent-sft
A clean, deduplicated, mixture-balanced tool-calling agent SFT dataset for fine-tuning google/gemma-4-26B-A4B-it, normalized from three agentic sources (Agent-Ark/Toucan-1.5M, open-thoughts/AgentTrove, nvidia/Nemotron-SFT-Agentic-v2).
Format (text, not pre-tokenized)
default config — 132,909 examples. Columns:
id, source, source_subset, tool_names (list)
messages — JSON string: list of {role, content, tool_calls, tool_responses}
tools — JSON… See the full description on the dataset page: https://huggingface.co/datasets/voidful/gemma4-agent-sft.blux-ca-dataset
BLUX cA Dataset
Deterministic dataset for BLUX-cA (Constitutional Agent) — a policy-driven validation and governance engine.
🔑 What This Dataset Is
This dataset is:
generated from a live deterministic engine (blux-ca)
verified against real acceptance runs
used to detect behavioral drift
Each row represents a fully validated engine execution.
🧠 Engine Mapping
Dataset Version
Engine
v1.0
cA-1.0-pro
This dataset is locked to the engine… See the full description on the dataset page: https://huggingface.co/datasets/Outer-Void/blux-ca-dataset.blux-coga-dataset
BLUX CogA Dataset
Deterministic dataset for BLUX-CogA (Cognitive Agent) — a structured reasoning engine.
🔑 What This Dataset Is
This dataset represents:
real outputs from the CogA engine
structured reasoning artifacts
deterministic reasoning verdicts
🧠 Engine Mapping
Dataset Version
Engine
v1.0
CogA-1.0-pro
📄 Data Format (JSONL)
Each row contains:
{
"problem": {...},
"thought_artifact": {...},
"reasoning_verdict": {...}… See the full description on the dataset page: https://huggingface.co/datasets/Outer-Void/blux-coga-dataset.nq-gar
Dataset Card for NQ-GAR
tw-pretrain