datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
corral_runs_reports
Corral – Evaluation Score Reports
Reports from Corral evaluation runs across models, scaffolds, scopes, and task granularities in all 8 environments
📋 Dataset Summary
This dataset is part of the Corral collection accompanying the paper AI scientists produce results without reasoning scientifically. It contains the Reports produced during the evaluation runs of models across all 8 Corral environments.
The dataset is organized into 24 configurations… See the full description on the dataset page: https://huggingface.co/datasets/jablonkagroup/corral_runs_reports.qwopus-dflash-swe20-runtime-results
Qwopus / DFlash SWE20 Runtime Results
Local RTX 3090 Ti benchmark artifacts for 20 long SWE-bench Lite prompts. The run compares Qwopus 3.6 GGUF variants, llama.cpp MTP speculative decoding, QuinsZouls, and DFlash DDTree configurations at 64K context with q8/q8 KV unless noted.
The quality score is a reproducible proxy rubric over gold-patch signals, not official SWE-bench pass/fail. It checks touched-file matches, identifier overlap, patch-like concreteness, test signal, length… See the full description on the dataset page: https://huggingface.co/datasets/jakeatx/qwopus-dflash-swe20-runtime-results.SciCode-Runnable-Benchmark-Reviewedrunning-coach-sft
Running Coach SFT
Instruction-tuning data for a distance-running coaching assistant. Every pace,
split, and race-equivalent in the corpus is computed from a Daniels/Gilbert VDOT
implementation rather than written into a template, so the numbers are internally
consistent across all 1,500 examples.
Why this exists
Coaching corpora scraped from forums and blogs teach a model the register of
coaching without the arithmetic underneath it. A model that interpolates… See the full description on the dataset page: https://huggingface.co/datasets/hoodarunner/running-coach-sft.runpod_qwen32_benign_thoughts_casestudies_rescuedtool-reasoning-sft-RESEARCH-grill-lab-browsecomp-plus-runs-data-cleaned-rectified
Tool-Reasoning SFT — BrowseComp-Plus Runs (Cleaned & Rectified)
Multi-turn tool-use reasoning trajectories derived from grill-lab/browsecomp-plus-runs, converted to a structured SFT format following the interstellarninja/hermes_reasoning_tool_use convention.
Source
Based on the execution trajectories from "Revisiting Text Ranking in Deep Research" (arXiv:2602.21456):
Original data: grill-lab/browsecomp-plus-runs (MIT)
Format
Each row contains a messages… See the full description on the dataset page: https://huggingface.co/datasets/AmanPriyanshu/tool-reasoning-sft-RESEARCH-grill-lab-browsecomp-plus-runs-data-cleaned-rectified.opentraces-runtime
opentraces-runtime
Community-contributed agent traces in opentraces JSONL format.
Usage
from datasets import load_dataset
ds = load_dataset("OpenTraces/opentraces-runtime")
Schema
Each JSONL line is a TraceRecord containing:
trace_id: Unique identifier for the trace
session_id: Source session identifier
agent: Agent identity (name, version, model)
task: Structured task metadata
steps: List of LLM API calls (thought-action-observation… See the full description on the dataset page: https://huggingface.co/datasets/OpenTraces/opentraces-runtime.gpt-5.4-frontend-development-11062026
GPT-5.4 Frontend Development Dataset (11062026)
This dataset is a synthetic chat-formatted code dataset focused on frontend development tasks in React and TypeScript.
It contains 1032 JSONL records collected on 2026-06-11 and generated with GPT-5.4 from frontend-oriented prompts covering reusable UI, compact feature units, forms, widgets, and related interface implementation tasks.
Overview
Each record contains:
task_id - numeric task identifier
category - task… See the full description on the dataset page: https://huggingface.co/datasets/runanlab/gpt-5.4-frontend-development-11062026.runux-tpu-v5e-benchmarks
⚡ RunuX-AI — TPU v5e Inference Benchmarks
Achieving 3× Throughput & 3× Energy Reduction on Google TPU v5e
Xavier Callens · Socrate AI Lab (Non-Profit)
Reproducible benchmark data & scripts — No proprietary code included
🎯 What Is This?
This repository contains benchmark results and Apache-2.0 reproduction scripts for comparing LLM inference performance across 5 frameworks on Google TPU v5e. The goal is to enable independent verification of our claims… See the full description on the dataset page: https://huggingface.co/datasets/callensxavier/runux-tpu-v5e-benchmarks.python-runtime-verified-error-correction
Python Runtime-Verified Error Correction Dataset 🐍⚡
Overview
Production-grade synthetic dataset of Python code errors with runtime-verified corrections. Each sample contains broken code, the actual runtime error, and a guaranteed-working fix validated through execution.
Unlike traditional synthetic datasets, every correction is verified by actually running the code in an isolated environment—eliminating hallucinations and ensuring real-world applicability.… See the full description on the dataset page: https://huggingface.co/datasets/SyntheticLogic-Labs/python-runtime-verified-error-correction.kakugo-run
Kakugo Rundi dataset
[Paper] [Code] [Model]
A synthetically generated conversation dataset for training in Rundi.
This dataset contains synthetic conversational data and translated instructions designed to train Small Language Models (SLMs) for Rundi. It was generated using the Kakugo pipeline, a method for distilling high-quality capabilities from a large teacher model into low-resource language models. The teacher model used to generate this dataset was… See the full description on the dataset page: https://huggingface.co/datasets/ptrdvn/kakugo-run.thai-sql-eval
thai-sql-eval
Text-to-SQL evaluation set for Thai (and English) natural-language questions against
real Thai government open data, across 4 domains: budget, demographics, education,
healthcare. Every question is authored in both Thai and English, mapped to the
same gold SQL, so you can directly compare a model's Thai vs. English text-to-SQL
ability. 300 rows total (150 question/SQL pairs × 2 languages).
Built because no public Thai text-to-SQL benchmark existed at the time — the… See the full description on the dataset page: https://huggingface.co/datasets/Rungrod/thai-sql-eval.SafeRun-RunPlanning-Benchmark
SafeRun Running-Planning Benchmark
A benchmark for evaluating deterministic, safety-constrained LLM planning in the
running-training domain. It accompanies the paper SafeRun: Enabling Determinism in
LLM Planning for Running.
Each sample asks a model to produce a multi-day/multi-week running training plan that
satisfies hard physiological safety rules while following the user's natural-language
request. The benchmark is designed to stress-test whether an LLM (or an LLM + tool… See the full description on the dataset page: https://huggingface.co/datasets/zzp-seeker/SafeRun-RunPlanning-Benchmark.hf-coding-tools-dashboard-run-april12
HuggingFace AI Coding Tools Dashboard
Benchmark data from the HuggingFace AI Dashboard — tracking how AI coding tools (Claude Code, Codex, Copilot, Cursor) recommend HuggingFace products across 32 developer categories.
Dataset Structure
Split
Description
Rows
results
Full benchmark results with LLM responses, cost, tokens, latency, and product detection
8875
queries
Benchmark query definitions across 32 categories
263
runs
Run metadata and tool/model… See the full description on the dataset page: https://huggingface.co/datasets/davidkling/hf-coding-tools-dashboard-run-april12.gpt-5.4-frontend-development-27052026
Site Coding Dataset
Site Coding Dataset is a synthetic chat-formatted dataset for code generation, focused on frontend development, UI implementation, and instruction-following programming tasks.
Site Coding Dataset — синтетический датасет в chat-формате для генерации кода, сфокусированный на frontend-разработке, UI-реализации и instruction-following задачах программирования.
Overview
This dataset contains 834 records in JSONL format.Each record includes:
category — task… See the full description on the dataset page: https://huggingface.co/datasets/runanlab/gpt-5.4-frontend-development-27052026.runpod_multi_model_think_content_casestudiesCode-170k-rundi
Dataset Description
Code-170k-rundi is a groundbreaking dataset containing 176,999 programming conversations, originally sourced from glaiveai/glaive-code-assistant-v2 and translated into Rundi, making coding education accessible to Rundi speakers.
🌟 Key Features
176,999 high-quality conversations about programming and coding
Pure Rundi language - democratizing coding education
Multi-turn dialogues covering various programming concepts
Diverse topics: algorithms, data… See the full description on the dataset page: https://huggingface.co/datasets/michsethowusu/Code-170k-rundi.cem-rung1-single-rare-traits
Rung-1 single-trait CEM optimization datasets
This private research dataset contains the judged optimization batches from the
rung-1 description+query CEM runs. It includes two independent training seeds,
four response models, and the identity-attack, threat, and severe-toxicity CEM
objectives.
Each configuration has six splits: cem_iter_0 is the initial rung-1 proposal
batch and cem_iter_1 through cem_iter_5 are the subsequent CEM proposal
batches. Each row retains the… See the full description on the dataset page: https://huggingface.co/datasets/singhalrk/cem-rung1-single-rare-traits.swe-explore-find-dev100-runs
SWE-Explore find — dev-100 evidence runs
Five complete 100-trial runs of the find (fault-localization) arm of SWE-Explore,
kept because each one is load-bearing evidence for a specific claim about the
scoring fixes on branch fix/swe-explore-find-scoring of harness_bench.
Every run is 100 trials of the same dev-100 find subset, run through
Harbor with the swe_explore.agents:PiSut agent. index.jsonl
has one row per trial (500 rows); the full raw Harbor trial directories are in… See the full description on the dataset page: https://huggingface.co/datasets/miaomiao64/swe-explore-find-dev100-runs.teacher-traces
Training Traces
Supplementary release for the paper Evaluating Agents Across Runtime Contracts: When Mismatch Costs Efficiency or Quality (IAEval 2026, the NeurIPS 2026 Workshop on Evaluation of Interactive Agents). This dataset
holds the teacher agent traces used to fine-tune the paper's LoRA adapters (see the
sibling cap-sweep-eval-data release and the fourteen adapter repos alongside this one).
6,000 agent traces total (1,000 per family x runtime combination), produced by an… See the full description on the dataset page: https://huggingface.co/datasets/runtime-contracts/teacher-traces.rung1-desc-query-psim-multiseed-50k
Rung-1 desc-query base-PSIM multiseed evaluation
This private dataset contains the completed 500,000-query rung-1 desc-query
experiment used for Monte Carlo/PSIM convergence analysis. It was generated
from the base PSIM desc-query model only; no CEM or QSIM samples are
included.
Organization
queries: 10 splits (seed_80, seed_82, ..., seed_98), each with
exactly 50,000 stop-finished queries.
deepseek-llama, llama-3.1-8b, nemotron, and qwen3-8b: matching judged… See the full description on the dataset page: https://huggingface.co/datasets/singhalrk/rung1-desc-query-psim-multiseed-50k.hf-coding-tools-dashboard-run-april22-v2
HuggingFace AI Coding Tools Dashboard
Benchmark data from the HuggingFace AI Dashboard — tracking how AI coding tools (Claude Code, Codex, Copilot, Cursor) recommend HuggingFace products across 32 developer categories.
Dataset Structure
Split
Description
Rows
results
Full benchmark results with LLM responses, cost, tokens, latency, and product detection
728
queries
Benchmark query definitions across 32 categories
141
runs
Run metadata and tool/model… See the full description on the dataset page: https://huggingface.co/datasets/davidkling/hf-coding-tools-dashboard-run-april22-v2.SceneChain-12K
SceneChain-12K
SceneChain-12K is a multi-turn scene editing conversation dataset for training vision-language models to generate and iteratively refine 3D indoor scenes.
Data Format
Each sample is a JSON object with:
messages: Multi-turn conversation following OpenAI chat format (system/user/assistant)
images: List of rendered scene image paths (relative to dataset root)
Conversation Structure
System: Scene editing instructions and tool definitions
User:… See the full description on the dataset page: https://huggingface.co/datasets/runder1/SceneChain-12K.agents-learn-runtime-train
Agents Learn Their Runtime -- Training Traces
Paper: Agents Learn Their Runtime: Interpreter Persistence as Training-Time Semantics
2,000 agent traces (1,000 with a persistent interpreter, 1,000 with a stateless interpreter) produced by Gemini 3 Flash solving Opaque Knapsack tasks (as defined in the paper). These traces were used to fine-tune two LoRA adapters on Qwen3-8B, one per runtime condition. The paper's central question is whether the execution semantics baked into training… See the full description on the dataset page: https://huggingface.co/datasets/pythonformer/agents-learn-runtime-train.delhIA-runs
delhIA — Conversation Runs
This dataset contains 150 conversation runs on the delhIA Météo-France domain:
50 evaluation scenarios × 3 LLM assistants, produced by a fully automated multi-turn simulation framework.
Conversations are stored in OpenAI Chat Completions format (roles: assistant, user, tool).
Dataset at a Glance
Runs
150 (50 tasks × 3 models)
Models
gpt-4o, qwen3-32b, qwen3-8b
Domain
Météo-France weather assistant
Language
French
Format
JSON… See the full description on the dataset page: https://huggingface.co/datasets/Jourdain/delhIA-runs.smolified-debug-run
🤏 smolified-debug-run
Intelligence, Distilled.
This is a synthetic training corpus generated by the Smolify Foundry.
It was used to train the corresponding model smolify/smolified-debug-run.
📦 Asset Details
Origin: Smolify Foundry (Job ID: DEBUG_RETRY)
Records: 200
Type: Synthetic Instruction Tuning Data
⚖️ License & Ownership
This dataset is a sovereign asset owned by smolify.
Generated via Smolify.ai.
elder-futhark-rune-meanings
Elder Futhark Rune Meanings — AskRunes Dataset
A structured dataset of all 24 Elder Futhark runes with divinatory meanings, curated by AskRunes.com — a free rune reading platform offering 22 curated spreads interpreted around the user's question.
Each record contains the rune's traditional attributes (position, Unicode symbol, transliteration, literal meaning, aett) and its full divinatory interpretations exported verbatim from AskRunes: general symbolism, upright and reversed… See the full description on the dataset page: https://huggingface.co/datasets/AskRunes/elder-futhark-rune-meanings.aurelia-runtime-corpus-v2
Aurelia Runtime Corpus v2
Aurelia Runtime Corpus v2 is a schema-gated validator corpus for the Aurelia causal world-simulation engine. It trains and evaluates a runtime-model coprocessor to distinguish verified mechanical/lore-grounded simulation records from subtly invalid candidates.
This release is the first full v2 local build: 25 fresh simulation seeds, 20 years per seed, all five Aurelia worlds, with grouped train/validation/test splits and balanced deterministic + AAA… See the full description on the dataset page: https://huggingface.co/datasets/OusiaResearch/aurelia-runtime-corpus-v2.