datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
python_code_instructions_18k_alpaca
Dataset Card for python_code_instructions_18k_alpaca
The dataset contains problem descriptions and code in python language.
This dataset is taken from sahil2801/code_instructions_120k, which adds a prompt column in alpaca style. Refer to the source here.
code_instructions_120k_alpaca
Dataset Card for code_instructions_120k_alpaca
This dataset is taken from sahil2801/code_instructions_120k, which adds a prompt column in alpaca style. Refer to the original source here.
code_contest_python3_alpaca
Dataset Card for Code Contest Processed
Dataset Summary
This dataset contains coding contest questions and their solution written in Python3.
This dataset is created by processing code_contest dataset from Deepmind. It is a competitive programming dataset for machine-learning. Read more about dataset at original source.
Columns Description
id : unique string associated with a problem
description : problem description
code : one correct code for the problem… See the full description on the dataset page: https://huggingface.co/datasets/iamtarun/code_contest_python3_alpaca.Fable-5-traces
Glint Research Dataset Card
Fable 5 Pi Agent Traces
A compact, high-signal corpus of Fable 5 coding-agent traces converted into Hugging Face Agent Traces / Pi-compatible sessions for Data Studio inspection, tool-use policy learning, and reasoning/action distillation.
Primary Config
pi_agent/train
Agent Trace preview enabled
4,665 Pi trace sessions
60 source sessions
3,799 tool… See the full description on the dataset page: https://huggingface.co/datasets/IAMRonHIT/Fable-5-traces.ui-instruct-4k
UI Instruct 4K
A instruction-completion dataset for finetuning language models to specialize in generating Next.js / ShadCN UI components using React, TypeScript, and Tailwind CSS.
Dataset Summary
This dataset was created with the primary goal of finetuning Qwen 3.5 4B to become a specialist at outputting production-ready Next.js and ShadCN-based UI components. Each example consists of a natural language prompt describing a UI component or layout, paired with a clean… See the full description on the dataset page: https://huggingface.co/datasets/iamdyeus/ui-instruct-4k.albedo_904k
albedo_904k
Merged, last-turn-cleaned SFT corpus of mini-swe-agent trajectories generated by
three strong teacher models. Each row is a multi-turn messages list; the
training target is the last assistant turn only.
Fields
messages: list of {role, content} turns (system / user / assistant ...).
model: teacher that generated the completion.
Composition (904,692 rows)
model
rows
Qwen3-Next
751,689
Kimi-K2.6
135,772
deepseek-v3.2
17,231… See the full description on the dataset page: https://huggingface.co/datasets/iamPi/albedo_904k.code_contest_processed
Dataset Card for Code Contest Processed
Dataset Summary
This dataset is created by processing code_contest dataset from Deepmind. It is a competitive programming dataset for machine-learning. Read more about dataset at original source.
Columns Description
id : unique string associated with a problem
description : problem description
code : one correct code for the problem
language : programming language used for code
test_samples : contains inputs and their… See the full description on the dataset page: https://huggingface.co/datasets/iamtarun/code_contest_processed.tessera-8a92b237
tessera — corpus epoch 13, full sweep
Teacher-anchored SFT data harvested from every published Affine (Bittensor
SN120) duel scored against corpus epoch 13 — 230 duel records, chal-00760
through chal-01102, covering 2026-08-16 to 2026-08-24.
42,006 rows over 42,006 distinct turns (one row per turn), drawn from
4,981 trajectories and 3,751 strata. That is 70% of the 59,745-turn epoch-13
corpus, and 2.3× the 18,138 rows of
iamPi/tessera-77d11909,
which sampled a subset of the same… See the full description on the dataset page: https://huggingface.co/datasets/iamPi/tessera-8a92b237.GPokeT2-dataset
🎮 GPokeT2 — Pokémon Sprite Dataset
Pokémon sprites from all mainline Gen 3 and Gen 4 games, encoded as ASCII token sequences and paired with rich metadata conditioning labels. Used to train GPokeT2.
Pokemon sprite
ASCII representation
Train the model
<img src="data:image/png;base64… See the full description on the dataset page: https://huggingface.co/datasets/iamthinbaker/GPokeT2-dataset.Complete-FABLE.5-traces-2M
Complete FABLE.5 Traces 2M
Full FABLE.5 / Mythos corpus restored, with session-limit answer rows removed.
Dataset Viewer | Parquet | Raw JSONL.gz
This dataset is a post-closure compilation of all available FABLE.5 / Mythos trace datasets found on Hugging Face during the curation pass after the closure of Fable and Mythos. It is deduplicated at the normalized-row level and keeps row-level provenance through first_source_dataset, first_source_config, first_source_split… See the full description on the dataset page: https://huggingface.co/datasets/IAMRonHIT/Complete-FABLE.5-traces-2M.japanese-humor-evaluation-v2
Japanese Multimodal Humor Evaluation Dataset (v2)
画像/テキストのお題に対する面白い回答のデータセット。bokete（画像→テキスト）とkeitai（テキスト→テキスト）を統合。
使い方
from datasets import load_dataset
dataset = load_dataset("iammytoo/japanese-humor-evaluation-v2")
データ構造
odai_type: 'image' or 'text'
image: 画像お題（textタイプではNone）
odai: テキストお題（imageタイプではNone）
response: 回答テキスト
score: 0-4の正規化スコア
ソース
YANS-official/ogiri-bokete
YANS-official/ogiri-keitai
nepali_news_textGlobal-Population-Data
List of Countries and Dependencies by Population
This dataset contains population-related information for countries and dependencies, scraped from Wikipedia. The dataset includes the following columns:
Location: The country or dependency name.
Population: Total population count.
% of World: The percentage of the world's population this country or dependency represents.
Date: The date of the population estimate.
Source: Whether the source is official or derived from the United… See the full description on the dataset page: https://huggingface.co/datasets/iamramzan/Global-Population-Data.gpt-oss-20b-reasoning-traces
GPT-OSS-20B Reasoning Traces
3,333 reasoning traces generated by openai/gpt-oss-20b and filtered for clean, terminating reasoning. It was built to distill GPT-OSS's tight reasoning style into smaller models, and is the training set behind iAmBoosted/Qwen3.5-9B-OSS-Distilled.
What's in it
Each record pairs a prompt with GPT-OSS-20B's full reasoning trace and final answer, in chat-message form, ready for supervised fine-tuning (SFT).
~4,000 raw traces were generated, then… See the full description on the dataset page: https://huggingface.co/datasets/iAmBoosted/gpt-oss-20b-reasoning-traces.execution-verified-agent-trajectories
Execution-Verified Agent Trajectories — Format & Method
This repository documents a method and data format for building supervised fine-tuning sets from agent
trajectories that are verified by running the code, not by asking a model whether the answer looks right.
This is a specification plus synthetic examples, not a corpus. The trajectories that trained
Luthor 8B were generated against a private repository and cannot be
released. Everything needed to rebuild an equivalent set… See the full description on the dataset page: https://huggingface.co/datasets/IAMIbrahim/execution-verified-agent-trajectories.aws_cross_account_iam_assume_role_chain_stall_teaser
🚀 Cloud Infrastructure - AWS Cross-Account IAM AssumeRole Chain Stall Triage (Evaluation Teaser)
⚡ Official Free Evaluation Teaser (50 Verified Multi-Turn Scenarios)🏆 Get the Full Production Package (500 Samples) & Commercial EULA on Gumroad:👉 Purchase Full Production Master Dataset on Gumroad🏷️ Use coupon code LAUNCH20 for 20 € off at checkout!
🌟 Domain Focus & Capabilities
Diagnoses STS AssumeRole rate limits, circular trust relationship stalls, and… See the full description on the dataset page: https://huggingface.co/datasets/emgena/aws_cross_account_iam_assume_role_chain_stall_teaser.Boxing-Champions
World Heavyweight Boxing Champions Dataset
This dataset contains information about world heavyweight boxing champions extracted from the Wikipedia page. It includes details such as champion names, reign periods, and title defenses.
Dataset Structure
Columns
Column Name
Description
No
The ordinal number of the champion.
Champion
Name of the heavyweight boxing champion.
Recognition
The organization or title under which the reign was recognized.… See the full description on the dataset page: https://huggingface.co/datasets/iamramzan/Boxing-Champions.MediFlowThinks
MediFlow
A large-scale synthetic instruction dataset of 2.5M rows (~700k unique instructions) for clinical natural language processing covering 14 task types and 98 fine-grained input clinical documents.
t-SNE 2D Plot of MediFlow Embeddings by Task Types
Dataset Splits
mediflow: 2.5M instruction data for SFT alignment.
mediflow_dpo: ~135k top-quality instructions with GPT-4o generated rejected_output for DPO alignment.
Main Columns
instruction:… See the full description on the dataset page: https://huggingface.co/datasets/IAMRonHIT/MediFlowThinks.orrery-ac54af81
halyard-ac54af81
Format-repaired SFT corpus for Bittensor SN120 (Affine), Reason v4
(weight_version_key=7). TRL prompt-completion format.
What was repaired
Every row's thought channel was normalised to exactly `
THOUGHT: {{z}}
{{y}}`.
40.08% of rows (7,270 / 18,138) carried at least one defect:
defect
rows
stray <think> / </think> / <thinking> / </thinking> inside the thought
7,124
corrupted THOUGHT label (THO, THOTH:, THOFTWARE:, THOTHought:, THOURLY… See the full description on the dataset page: https://huggingface.co/datasets/iamPi/orrery-ac54af81.Largest-Banks
Dataset Summary
This dataset contains information about the largest banks globally, including their rank, name, and total assets (in US$ billion as of 2023). The data was scraped from Wikipedia's List of Largest Banks. It can be used for financial analysis, market research, and educational purposes.
Dataset Structure
Columns
Rank: The rank of the bank based on total assets.
Bank Name: The name of the bank.
Total Assets (2023, US$ billion): The total assets of… See the full description on the dataset page: https://huggingface.co/datasets/iamramzan/Largest-Banks.Self-Instruct-Japanese-Elzya-13BA Japanese dataset generated with an opensource elyza/ELYZA-japanese-Llama-2-13b-instruct model.
This dataset is used in evaluating AI-generated text detection methods and is well-suited for self-instruct methods.
The instructions were taken from:
https://huggingface.co/datasets/CausalLM/GPT-4-Self-Instruct-Japanese
The model used is:
https://huggingface.co/elyza/ELYZA-japanese-Llama-2-7b
License: refer to the model's license
verdigris-d2d8b9d2
verdigris-d2d8b9d2
Cleaned SFT corpus for Bittensor SN120 (Affine), Reason v4
(weight_version_key=7). 17,655 rows, TRL prompt-completion format.
Format is exactly `
THOUGHT: {z}
{y}` on every row.
How it was cleaned
Derived from iamPi/tessera-77d11909.
Prompts and actions are byte-identical to the source; only the thought
channel changed, and 483 rows (2.66%) were removed.
Repaired in place (40.08% of rows had at least one defect)
defect
rows… See the full description on the dataset page: https://huggingface.co/datasets/iamPi/verdigris-d2d8b9d2.sabaton-914184bb
sabaton-914184bb
Cleaned SFT corpus for Bittensor SN120 (Affine), Reason v4
(weight_version_key=7). 17,655 rows, TRL prompt-completion format.
Format is exactly `
THOUGHT: {z}
{y}` on every row.
How it was cleaned
Derived from iamPi/tessera-77d11909.
Prompts and actions are byte-identical to the source; only the thought
channel changed, and 483 rows (2.66%) were removed.
Repaired in place (40.08% of rows had at least one defect)
defect
rows… See the full description on the dataset page: https://huggingface.co/datasets/iamPi/sabaton-914184bb.qa_metacul
qa_metacul
Summary
qa_metacul is an 800-question multiple-choice benchmark used to evaluate metadata-conditioned language models in the Metadata Conditioned LLMs project.
The benchmark tests whether a model can answer culturally and geographically grounded factual questions for different parts of the world, and whether metadata-aware models correctly adapt their answers when continent- or country-level context changes.
Paper: https://arxiv.org/abs/2601.15236
Project… See the full description on the dataset page: https://huggingface.co/datasets/iamshnoo/qa_metacul.ai-basic-law-dataset
台灣人工智慧基本法 訓練資料集
Taiwan AI Basic Law (人工智慧基本法) Q&A dataset for LLM finetuning.
Files
File
Description
Entries
train.jsonl
Full training dataset with oversampling
~5000
fulltext.jsonl
Clean article fulltext (20 articles)
38
Data Composition
Category
Unique
Repeat
Purpose
Article Fulltext Q&A
~157
x15
Verbatim article text with topic anchors
Alias Recognition
~109
x10
「基本法」「AI基本法」→ 人工智慧基本法
Legislative Reasons
~35
x3
Background… See the full description on the dataset page: https://huggingface.co/datasets/iamjry/ai-basic-law-dataset.mental_health_Chatbot
Amod/mental_health_counseling_conversations
This dataset is a compilation of high-quality, real one-on-one mental health counseling conversations between individuals and licensed professionals. Each exchange is structured as a clear question–answer pair, making it directly suitable for fine-tuning or instruction-tuning language models that need to handle sensitive, empathetic, and contextually aware dialogue.
Since its public release in 2023, it has been downloaded over 100,000… See the full description on the dataset page: https://huggingface.co/datasets/Iamzoo/mental_health_Chatbot.halyard-9b1a09e7
halyard-9b1a09e7
Format-repaired SFT corpus for Bittensor SN120 (Affine), Reason v4
(weight_version_key=7). TRL prompt-completion format.
What was repaired
Every row's thought channel was normalised to exactly `
THOUGHT: {{z}}
{{y}}`.
40.08% of rows (7,270 / 18,138) carried at least one defect:
defect
rows
stray <think> / </think> / <thinking> / </thinking> inside the thought
7,124
corrupted THOUGHT label (THO, THOTH:, THOFTWARE:, THOTHought:, THOURLY… See the full description on the dataset page: https://huggingface.co/datasets/iamPi/halyard-9b1a09e7.HCMUT_FAQpython_code_instructions_18k_alpaca
Dataset Card for python_code_instructions_18k_alpaca
The dataset contains problem descriptions and code in python language.
This dataset is taken from sahil2801/code_instructions_120k, which adds a prompt column in alpaca style. Refer to the source here.
orpo-dpo-mix-40k
ORPO-DPO-mix-40k v1.2
This dataset is designed for ORPO or DPO training.
See Fine-tune Llama 3 with ORPO for more information about how to use it.
It is a combination of the following high-quality DPO datasets:
argilla/Capybara-Preferences: highly scored chosen answers >=5 (7,424 samples)
argilla/distilabel-intel-orca-dpo-pairs: highly scored chosen answers >=9, not in GSM8K (2,299 samples)
argilla/ultrafeedback-binarized-preferences-cleaned: highly scored chosen answers >=5… See the full description on the dataset page: https://huggingface.co/datasets/Iambackup/orpo-dpo-mix-40k.