Theme
AI Resources
General365
General365 is a manually curated benchmark for evaluating general reasoning in LLMs across difficult and diverse tasks, with a focus on reasoning over K-12-scope knowledge rather than domain-specialist knowledge.
The official repository presents General365 as the benchmark release for the paper on general reasoning under high difficulty and diversity, with public questions, variants, model-response formatting, grading code, project links, leaderboard materials, and a Hugging Face dataset link. Use this as a first read, not a recommendation. Open the original project before trusting details like terms, limits, privacy, cost, setup, or safety.
What it is
A general-reasoning benchmark
General365 is framed as a benchmark for testing broad reasoning ability, with manually crafted seed problems and variants intended to reduce overreliance on narrow domain knowledge or rote memorization.
Why it stands out
Difficulty and diversity focus
The notable angle is the combination of high-difficulty tasks, broad scenario coverage, K-12-scope knowledge constraints, held-out questions, and hybrid scoring that mixes rule-based and model-based checks.
Availability
Repo, dataset, project page, and leaderboard
The official materials include the GitHub repository, paper link, project page, leaderboard, Hugging Face dataset, grading script, model-response format, and example workflow for running evaluations.
Why it matters
Why readers may notice it
Reasoning benchmarks can blur into knowledge tests. This project is positioned around separating general reasoning skill from specialist academic knowledge, while keeping tasks difficult and varied.
What readers may want to know
Where it fits
Open it as part of the benchmark and dataset layer rather than the model, app, or agent layer. It is most relevant to readers comparing model reasoning evaluations and benchmark design choices.
Reporting note
What appears notable
The repository is useful for checking the manually curated question design, 365 seed problems, 1,095 variants, held-out test-set note, leaderboard materials, and hybrid scoring workflow.
Before using
What readers may want to review
Which public questions, variants, and held-out limitations are described in the official materials.
The grading script, model-response JSONL format, and scoring method before adapting the benchmark.
Whether the benchmark is being used to compare general reasoning, data contamination risk, or a narrower model capability claim.
Reader fit
Who may find it relevant
Readers following LLM reasoning benchmarks and model-comparison methods.
Builders and researchers comparing difficult general-reasoning tasks across model families.
Less relevant for readers looking for a model checkpoint, finished app, or agent framework.
Editorial note
Why it is included here
Use General365 as a source check on reasoning evaluation across varied and difficult tasks.
Source links
Original materials
Reader note
Before relying on this entry
LifeHubber lists entries to help readers inspect AI projects, not to endorse them or prove they are safe, suitable, accurate, maintained, or right for a specific use. We do not verify every entry in depth. Before relying on anything listed, review the original materials, terms, privacy practices, limits, and risks that matter for your situation.
More in Datasets
Keep browsing this category
A few more places to continue in datasets.
ClawMark
evolvent-ai/ClawMark
A living-world benchmark for multi-day, multimodal coworker agents, spanning 100 tasks across professional domains and real tool environments.
LARYBench
meituan-longcat/LARYBench
A benchmark for evaluating latent action representations, with pipelines for action semantics, robotic control regression, and broader vision-to-action alignment.
Monitorability Evals
openai/monitorability-evals
An OpenAI evaluation-data release for studying monitorability, with public eval splits, prompt templates, dataset mappings, and metric code from the Monitoring Monitorability paper.
Related in LifeHubber
Keep the thread going
Follow the next layer with AI Resources for AI projects worth inspecting at the source, AI Guides for decision habits for messy AI choices, AI Access for free and low-cost ways to compare AI model access, AI Ballot for a clearer view of what readers are leaning toward, and AI Radar for AI stories that deserve a second look.