datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
code-search-net-python
Dataset Card for "code-search-net-python"
Dataset Description
Homepage: None
Repository: https://huggingface.co/datasets/Nan-Do/code-search-net-python
Paper: None
Leaderboard: None
Point of Contact: @Nan-Do
Dataset Summary
This dataset is the Python portion of the CodeSarchNet annotated with a summary column.The code-search-net dataset includes open source functions that include comments found at GitHub.The summary is a short description of what the… See the full description on the dataset page: https://huggingface.co/datasets/Nan-Do/code-search-net-python.nanochat-climbmix-arithmetic-base10
nanochat ClimbMix + Base-10 Arithmetic
This dataset contains the first 170 shuffled ClimbMix training shards
used by nanochat's speedrun. The deterministic base-10 arithmetic corpus is
mixed into shards 00000..00149; the final
20 train shards are unchanged web-only padding.
The original validation shard (shard_06542.parquet) is also
copied unchanged.
Arithmetic corpus
Family
Examples
a + b = c (all ordered pairs 0..2000, two exposures)
8,008,002
a + b… See the full description on the dataset page: https://huggingface.co/datasets/Yujivus/nanochat-climbmix-arithmetic-base10.nanochat-climbmix-arithmetic-base7
nanochat ClimbMix + Arithmetic: base-7 numeral world
This is a deterministic base-7 rendering of
Yujivus/nanochat-climbmix-arithmetic-base10. It preserves
the exact shard names, row order, document order, arithmetic-document placement,
and non-numeric text of the source dataset.
Transformation rule
Every maximal ASCII digit run matching [0-9]+ is interpreted as a base-10
integer and rendered in base 7. Leading zeros are preserved as a prefix; signs,
punctuation… See the full description on the dataset page: https://huggingface.co/datasets/Yujivus/nanochat-climbmix-arithmetic-base7.instructional_code-search-net-python
Dataset Card for "instructional_code-search-net-python"
Dataset Summary
This is an instructional dataset for Python.
The dataset contains two different kind of tasks:
Given a piece of code generate a description of what it does.
Given a description generate a piece of code that fulfils the description.
Languages
The dataset is in English.
Data Splits
There are no splits.
Dataset Creation
May of 2023
Curation Rationale
This… See the full description on the dataset page: https://huggingface.co/datasets/Nan-Do/instructional_code-search-net-python.nanochat-climbmix-170
nanochat ClimbMix: first 170 train shards
Convenience mirror of the exact initial ClimbMix slice downloaded by
python -m nanochat.dataset -n 170.
Contents
Training: shard_00000.parquet through shard_00169.parquet
Validation: shard_06542.parquet
manifest.json: pinned source revision, file list, and byte sizes
The Parquet shards are copied without modifying their rows or text.
Attribution and provenance
nanochat:… See the full description on the dataset page: https://huggingface.co/datasets/Yujivus/nanochat-climbmix-170.Pluto-Nano-1.0-Pretrain-v2
ASTRAI Pluto Nano 1.0 — Pretrain Mix (v2)
Curated multilingual pretraining corpus (~50 GB parquet, ~12 B tokens after tokenization) used for ASTRAI Pluto Nano 1.0, a 1 B-total / 50 M-active MoE model with 64 k vocabulary and 5 target languages (EN, PT, ES, ZH, HI).
v2 additions vs v1: OpenThoughts3 (CoT reasoning), openstax textbooks + peS2o (science), and reweighting for better balance. NOTE: factsense (openbmb) was used at training time but is not redistributed here due to its… See the full description on the dataset page: https://huggingface.co/datasets/ASTRAI-labs/Pluto-Nano-1.0-Pretrain-v2.nanochat-climbmix-arithmetic-base6
nanochat ClimbMix + Arithmetic: base-6 numeral world
This is a deterministic base-6 rendering of
Yujivus/nanochat-climbmix-arithmetic-base10. It preserves
the exact shard names, row order, document order, arithmetic-document placement,
and non-numeric text of the source dataset.
Transformation rule
Every maximal ASCII digit run matching [0-9]+ is interpreted as a base-10
integer and rendered in base 6. Leading zeros are preserved as a prefix; signs,
punctuation… See the full description on the dataset page: https://huggingface.co/datasets/Yujivus/nanochat-climbmix-arithmetic-base6.nanoswe-trajs-260812
nanoswe SWE-agent trajectories (v0)
A consolidation of the SWE-bench-style coding-agent trajectory corpora used to train
the nanoswe speedrun models. Each row is one
multi-turn agent trajectory (issue → tool-using rollout → patch), stored untokenized.
1,582,701 trajectories, 34 parquet shards, content-deduplicated on traj_hash.
Seed corpus: ricdomolm/mini-coder-trajs-400k;
the rest are derived SWE-smith / openhands / swe-zero conversions.
Schema
column… See the full description on the dataset page: https://huggingface.co/datasets/nanoswe/nanoswe-trajs-260812.local-llm-benchmark
Local LLM Benchmark — Technical and Uncensored Behavior (NVIDIA RTX 5070 Ti 16GB)
English | 简体中文 | 繁體中文 | 한국어 | Español | 日本語 | हिन्दी | Русский | Português | తెలుగు | Français | Deutsch | Italiano | Tiếng Việt | العربية | اردو | বাংলা | فارسی | Română | Türkçe
Manual evaluation results of local GGUF model variants on a single consumer machine,
combining two fully independent benchmarks:
technical/
uncensored/
Measures
capability: coding, systems, networking, DB, agents… See the full description on the dataset page: https://huggingface.co/datasets/nanimani/local-llm-benchmark.nanochat-jp-pretrain
nanochat-jp-pretrain
nanochat の日本語フォーク nanochat-jp で使用する 事前学習用日本語コーパス です．
LLM によるクリーニングを施した日本語ウェブテキストと，llm-jp の公開コーパスを混合したものを，nanochat のデータローダがそのまま読める parquet 形式で配布しています．
構成
以下の4つのソースを混合し，全体をシャッフルしています．
ソース
llm-jp-corpus-v4 の ja_fineweb-2 サブセット（後述の追加データクリーニングを適用）
llm-jp-corpus-midtraining-v2 ja/llm-jp-IPT_v0.3.2/ja_general.jsonl.gz
llm-jp-corpus-midtraining-v2 ja/llm-jp-IPT_v0.3.2/ja_reasoning.jsonl.gz
llm-jp/scaling-data-constrained-llms… See the full description on the dataset page: https://huggingface.co/datasets/tohoku-nlp/nanochat-jp-pretrain.swesmith-qwen3.6-35b-a3b
SWE-smith trajectories from Qwen3.6-35B-A3B
Multi-turn coding-agent trajectories (issue → tool-using rollout → patch) produced by
Qwen3.6-35B-A3B on SWE-smith tasks, stored untokenized.
This is the exact SFT corpus used for the harbor arm of the
nanoswe teacher-distillation experiments.
101,901 trajectories over 45,242 unique SWE-smith task instances (3 sampled rollouts
per task, ~2.25 surviving filtering), 53 parquet shards, ~1.4 GB.
≈1.96B training tokens = exactly one epoch… See the full description on the dataset page: https://huggingface.co/datasets/nanoswe/swesmith-qwen3.6-35b-a3b.code-search-net-java
Dataset Card for "code-search-net-java"
Dataset Summary
This dataset is the Java portion of the CodeSarchNet annotated with a summary column.The code-search-net dataset includes open source functions that include comments found at GitHub.The summary is a short description of what the function does.
Languages
The dataset's comments are in English and the functions are coded in Java
Data Splits
Train, test, validation labels are included in the… See the full description on the dataset page: https://huggingface.co/datasets/Nan-Do/code-search-net-java.nemotron-nano-30b-miniswe-swebench-verified
Nemotron Nano 30B + mini-swe-agent SWE-bench Verified Trajectories
Agent trajectories from running NVIDIA Nemotron 3 Nano 30B A3B (MoE, 8B active params) on SWE-bench Verified using mini-swe-agent.
⚠️ Incomplete Run
This benchmark was terminated early due to poor performance. The model struggled with the agentic coding task.
Model Information
Attribute
Value
Model
NVIDIA Nemotron 3 Nano 30B A3B
Architecture
MoE (30B total, 8B active)
Serving
vLLM… See the full description on the dataset page: https://huggingface.co/datasets/pankajmathur/nemotron-nano-30b-miniswe-swebench-verified.Medical-Reasoning-SFT-Nemotron-Nano-30B
Medical-Reasoning-SFT-Nemotron-Nano-30B
A large-scale medical reasoning dataset generated using nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16, containing over 444,000 samples with detailed chain-of-thought reasoning for medical and healthcare questions.
Dataset Overview
Metric
Value
Model
nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16
Total Samples
444,544
Samples with Reasoning
444,544 (100%)
Estimated Tokens
~1.01 Billion
Content Tokens
~808 Million… See the full description on the dataset page: https://huggingface.co/datasets/OpenMed/Medical-Reasoning-SFT-Nemotron-Nano-30B.nanochat-calendar-arithmetic-base10
nanochat Base-10 Calendar Arithmetic
A deterministic, base-10 arithmetic corpus scoped to three cyclic calendar
units: hour-of-day (mod 24), day-of-week (mod 7), and
month-of-year (mod 12). Companion to
Yujivus/nanochat-climbmix-arithmetic-base10,
built the same way but scoped to real modular calendar units instead of
free-integer add/sub/mul/div/mod.
Every example is a single line — question and answer collapsed into one
equation, no exposed reasoning:
23:00 + 18965h = 04:00… See the full description on the dataset page: https://huggingface.co/datasets/liodon-ai/nanochat-calendar-arithmetic-base10.nanoswe-trajs-v0
nanoswe SWE-agent trajectories (v0)
A consolidation of the SWE-bench-style coding-agent trajectory corpora used to train
the nanoswe speedrun models. Each row is one
multi-turn agent trajectory (issue → tool-using rollout → patch), stored untokenized.
1,582,701 trajectories, 34 parquet shards, content-deduplicated on traj_hash.
Seed corpus: ricdomolm/mini-coder-trajs-400k;
the rest are derived SWE-smith / openhands / swe-zero conversions.
Schema
column… See the full description on the dataset page: https://huggingface.co/datasets/ricdomolm/nanoswe-trajs-v0.NanoData
Dataset Description
To facilitate researchers to use NanoLM for comparative analysis across different model designs, we build a curated pre-training dataset from those of existing large-scale models (i.e., Llama, Falcon, GPT-3). It covers diverse domains to improve the generalization capabilities of the resultant models.
Dataset Creation
The data is mainly post-processed and filtered from RedPajama and RedPajamaV2.
We develop a series of cleaning steps to remove redundant… See the full description on the dataset page: https://huggingface.co/datasets/CofeAI/NanoData.SSR-RCoT-16K
SSR-RCoT-16K: Turning answer-only data into high-quality reasoning-supervision data
SSR-RCoT-16K is a public 16k subset derived from the data construction pipeline introduced in
Measuring and Mitigating Post-hoc Rationalization in Reverse Chain-of-Thought Generation.
This dataset is designed for powerful reasoning on general tasks, especially for the realistic setting where high-quality responses are available but chain-of-thought annotations are missing.
In such answer-rich but… See the full description on the dataset page: https://huggingface.co/datasets/Nanbeige/SSR-RCoT-16K.sakthai-combined-v7
SakThai Combined v7
Curated, larger-scale instruction-tuning data for tool-calling, function-calling, and agent-style reasoning in the SakThai model family.
Dataset Summary
SakThai Combined v7 extends the v6 family with more multi-turn examples, broader tool coverage, and stronger <tool>/function-calling formatting. It is intended for fine-tuning models that should invoke tools naturally, then continue the conversation after tool results.
Data Fields… See the full description on the dataset page: https://huggingface.co/datasets/Nanthasit/sakthai-combined-v7.atcoder_cot
Dataset Card for Atcoder-CoT
Dataset Description
Atcoder-CoT is a proof-of-concept dataset designed to demonstrate how a dataset like the one found here can be used to generate synthetic datasets for training reasoning models, particularly for Supervised Fine-Tuning (SFT) and Knowledge Distillation. It leverages human-created and debugged solutions, combined with LLM-generated text to create conversational turns. The approach can also be easily adapted to simulate human… See the full description on the dataset page: https://huggingface.co/datasets/Nan-Do/atcoder_cot.ManipuriGPT-Corpus-v1.0
ManipuriGPT Corpus v1.0
ManipuriGPT Corpus v1.0 is a research-grade, multi-script, deduplicated, and quality-scored corpus specifically engineered for pretraining Manipuri (Meiteilon) language foundation models.
Quick Summary
Total Sequences: 147,956
Total Tokens (ManipuriGPT-Tokenizer-v1.0): 4,347,075
Total Characters: 16,019,401
Pipeline Version: 5.6
Release Version: v1.0.0
Build Timestamp: 2026-07-25T09:17:21.960438Z
Primary Writing Systems… See the full description on the dataset page: https://huggingface.co/datasets/nanskong/ManipuriGPT-Corpus-v1.0.code-search-net-javascript
Dataset Card for "code-search-net-javascript"
Dataset Summary
This dataset is the JavaScript portion of the CodeSarchNet annotated with a summary column.The code-search-net dataset includes open source functions that include comments found at GitHub.The summary is a short description of what the function does.
Languages
The dataset's comments are in English and the functions are coded in JavaScript
Data Splits
Train, test, validation labels are… See the full description on the dataset page: https://huggingface.co/datasets/Nan-Do/code-search-net-javascript.nemotron-nano-hermes-traces
Nemotron Nano Hermes Agent Reasoning Traces
A curated dataset of reasoning traces for training local AI orchestrator agents.
Designed for SFT and RL training of Nemotron 3 Nano Omni to be the best local
Hermes Agent model.
Dataset Summary
Total SFT rows: 28,000
Total RL prompts: 28,000
Format: ShareGPT (conversations column)
Target model: nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16
Training framework: Unsloth Studio
Sources
Source
Count… See the full description on the dataset page: https://huggingface.co/datasets/r0b0tlab/nemotron-nano-hermes-traces.qualc-fineweb-edu-en
QualC FineWeb-Edu English (Cleaned)
QualC FineWeb-Edu English (Cleaned) is a cleaned subset of the official FineWeb-Edu dataset published by Hugging Face.
The dataset is intended for Large Language Model (LLM) pretraining, tokenizer training, continual pretraining, educational NLP research, and language modeling.
This repository contains approximately one million cleaned educational English documents prepared for the QualC project.
Dataset Information
Item… See the full description on the dataset page: https://huggingface.co/datasets/nandhakumarms/qualc-fineweb-edu-en.reason_code-search-net-python
Dataset Card for "reason_code-search-net-python"
Dataset Summary
This dataset is an instructional dataset for Python.The dataset contains five different kind of tasks.
Given a Python 3 function:
Type 1: Generate a summary explaining what it does. (For example: This function counts the number of objects stored in the jsonl file passed as input.)
Type 2: Generate a summary explaining what its input parameters represent ("For example: infile: a file descriptor of a file… See the full description on the dataset page: https://huggingface.co/datasets/Nan-Do/reason_code-search-net-python.qualc-fineweb-en
QualC FineWeb English (Cleaned)
QualC FineWeb English (Cleaned) is a cleaned English corpus derived from the FineWeb dataset published by Hugging Face.
The dataset is intended for Large Language Model (LLM) pretraining, tokenizer training, continual pretraining, and NLP research.
This repository contains a cleaned subset of the original FineWeb corpus collected and processed for the QualC project.
Dataset Summary
Language: English
Documents: ~1,000,000
Dataset… See the full description on the dataset page: https://huggingface.co/datasets/nandhakumarms/qualc-fineweb-en.ctf-dataset
ctf-dataset
CTF 与网络安全知识的 ShareGPT/ChatML 风格 SFT 数据集，可用于 LoRA 微调。
数据格式
每行是一个 JSON 对象，核心字段如下：
字段
说明
id
样本唯一 ID
dataset
数据集名称，当前为 ctf-dataset
category
来源主题或 CTF/安全类别
ctf_task_type
任务类型标签
messages
ShareGPT 消息数组，包含 system / user / assistant
metadata
来源路径、章节、字符数、chunk 等溯源信息
LLaMA-Factory 接入
将 ctf-dataset.jsonl 放入 LLaMA-Factory 的 data/ 目录后，在 data/dataset_info.json 中添加：
{
"ctf_dataset": {
"file_name": "ctf-dataset.jsonl"… See the full description on the dataset page: https://huggingface.co/datasets/Nanhang/ctf-dataset.openpii-masking-nano-1k
OpenPII Nano: Multilingual PII Masking Sample
A nano-sized stratified sample of OpenPII 1.5M,
perfect for quick prototyping, smoke tests, and CI fixtures. Every locale and every
label that exists in the parent dataset is represented in proportion.
📖 More information: www.ai4privacy.com/datasets/pii-masking-3m-asia-pacific
Dataset Details
Total Examples
Train
Validation
Labels
Languages
Regions
Annotations
Format
License
1,000
900
100
19
30
37
7… See the full description on the dataset page: https://huggingface.co/datasets/ai4privacy/openpii-masking-nano-1k.DiLLM.nano
DiLLM Phase 1 — Persian Corpora
The training corpora for DiLLM (دیبا),
a Persian-first language model built in Iran by Dibachain.
Publisher
Dibachain — dibachain.ir
Country
Iran 🇮🇷
Model
DibaAi/DiLLM.nano
Source code
github.com/AliAkrami1375/DibaLLM
Language
Persian (فارسی)
Total
~1.02M documents · ~379M tokens
This dataset was assembled from scratch for DiLLM. Every corpus was
downloaded from its primary source, filtered, and cleaned by the… See the full description on the dataset page: https://huggingface.co/datasets/DibaAi/DiLLM.nano.Nano3D-Edit-100k
Nano3D-Edit-100k
This dataset is the official data release for Nano3D, a training-free framework for precise and coherent 3D object editing without masks.
Paper: Nano3D: A Training-Free Approach for Efficient 3D Editing Without MasksProject Page: https://jamesyjl.github.io/Nano3D/
Nano3D integrates FlowEdit into TRELLIS to perform localized 3D edits guided by front-view renderings, and introduces Voxel/Slat-Merge strategies to preserve structural consistency between edited and… See the full description on the dataset page: https://huggingface.co/datasets/yejunliang23/Nano3D-Edit-100k.