datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
DeepSeek-v4-Pro-AgentThis dataset was generated using teich by TeichAI
Prepare these datasets for supervised fine-tuning in just a few lines of code — see the Conversion section below.
DeepSeek v4 Pro Agent Traces
This directory contains raw agent trace files generated by teich.
All assistant responses were generated by deepseek/deepseek-v4-pro.
JSONL files: 4006
Training-ready tools
A complete configured tools schema snapshot is embedded in the collapsed section at the bottom of… See the full description on the dataset page: https://huggingface.co/datasets/TeichAI/DeepSeek-v4-Pro-Agent.AM-DeepSeek-Distilled-40MFor more open-source datasets, models, and methodologies, please visit our GitHub repository and paper: DeepDistill: Enhancing LLM Reasoning Capabilities via Large-Scale Difficulty-Graded Data Training.
Due to certain constraints, we are only able to open-source a subset of the complete dataset.
Model Training Performance based on our complete dataset
On AIME 2024, our 72B model achieved a score of 79.2 using only supervised fine-tuning (SFT). The 32B model reached 75.8 and… See the full description on the dataset page: https://huggingface.co/datasets/a-m-team/AM-DeepSeek-Distilled-40M.DeepSeek-v4-Pro-AgentThis dataset was generated using teich by TeichAI
Prepare these datasets for supervised fine-tuning in just a few lines of code — see the Conversion section below.
DeepSeek v4 Pro Agent Traces
This directory contains raw agent trace files generated by teich.
All assistant responses were generated by deepseek/deepseek-v4-pro.
JSONL files: 4006
Training-ready tools
A complete configured tools schema snapshot is embedded in the collapsed section at the bottom of… See the full description on the dataset page: https://huggingface.co/datasets/ronaldcmz/DeepSeek-v4-Pro-Agent.deepseek-v4-pro-0813-agentic
DeepSeek-V4-Pro 0813 Agentic (DS4)
A standalone, verifiable-first agentic training corpus: 19,072 training traces
plus 2,135 held-out evaluation rows (validation 1,070 / test 1,065), generated by
DeepSeek-V4-Pro 0813 (deepseek-v4-pro-0813, official API, thinking mode) across 13 verifiable task families,
each row admitted only after passing a deterministic programmatic verifier. The corpus is
designed to be directly usable for SFT, GRPO/RLVR, and NeMo Gym / NeMo RL
(verified… See the full description on the dataset page: https://huggingface.co/datasets/r0b0tlab/deepseek-v4-pro-0813-agentic.Chinese-DeepSeek-R1-Distill-data-110k
中文基于满血DeepSeek-R1蒸馏数据集（Chinese-Data-Distill-From-R1）
🤗 Hugging Face | 🤖 ModelScope | 🚀 Github | 📑 Blog
注意：提供了直接SFT使用的版本，点击下载。将数据中的思考和答案整合成output字段，大部分SFT代码框架均可直接直接加载训练。
本数据集为中文开源蒸馏满血R1的数据集，数据集中不仅包含math数据，还包括大量的通用类型数据，总数量为110K。
为什么开源这个数据？
R1的效果十分强大，并且基于R1蒸馏数据SFT的小模型也展现出了强大的效果，但检索发现，大部分开源的R1蒸馏数据集均为英文数据集。 同时，R1的报告中展示，蒸馏模型中同时也使用了部分通用场景数据集。
为了帮助大家更好地复现R1蒸馏模型的效果，特此开源中文数据集。
该中文数据集中的数据分布如下：… See the full description on the dataset page: https://huggingface.co/datasets/Congliu/Chinese-DeepSeek-R1-Distill-data-110k.DeepSeek-v4-Pro-AgentThis dataset was generated using teich by TeichAI
Prepare these datasets for supervised fine-tuning in just a few lines of code — see the Conversion section below.
DeepSeek v4 Pro Agent Traces
This directory contains raw agent trace files generated by teich.
All assistant responses were generated by deepseek/deepseek-v4-pro.
JSONL files: 4006
Training-ready tools
A complete configured tools schema snapshot is embedded in the collapsed section at the bottom of… See the full description on the dataset page: https://huggingface.co/datasets/hardcoremoore/DeepSeek-v4-Pro-Agent.deepseek-hermes-reasoning-traces
DeepSeek V4 Pro Hermes Reasoning Traces
19,331 multi-turn ChatML + Hermes reasoning traces generated by DeepSeek V4 Pro. Designed for LoRA fine-tuning local models to operate as Hermes Agent instances.
Quick Start
\
Splits
Split
Traces
train
16,431
valid
1,933
test
967
Variants (VRAM-Tiered)
Variant
Max Tokens
Traces
GPU
nano
2,048
15,948
Dev / 7B
budget
4,096
2,149
48GB
standard
8,192
990
64GB
spark
16,384
244… See the full description on the dataset page: https://huggingface.co/datasets/r0b0tlab/deepseek-hermes-reasoning-traces.Chinese-DeepSeek-R1-Distill-data-110k-SFT
中文基于满血DeepSeek-R1蒸馏数据集（Chinese-Data-Distill-From-R1）
🤗 Hugging Face | 🤖 ModelScope | 🚀 Github | 📑 Blog
注意：该版本为，可以直接SFT使用的版本，将原始数据中的思考和答案整合成output字段，大部分SFT代码框架均可直接直接加载训练。
本数据集为中文开源蒸馏满血R1的数据集，数据集中不仅包含math数据，还包括大量的通用类型数据，总数量为110K。
为什么开源这个数据？
R1的效果十分强大，并且基于R1蒸馏数据SFT的小模型也展现出了强大的效果，但检索发现，大部分开源的R1蒸馏数据集均为英文数据集。 同时，R1的报告中展示，蒸馏模型中同时也使用了部分通用场景数据集。
为了帮助大家更好地复现R1蒸馏模型的效果，特此开源中文数据集。该中文数据集中的数据分布如下：
Math：共计36568个样本，
Exam：共计2432个样本，
STEM：共计12648个样本，… See the full description on the dataset page: https://huggingface.co/datasets/Congliu/Chinese-DeepSeek-R1-Distill-data-110k-SFT.deepseek-v4-tiny-cpu-repro-v1
DeepSeek-V4 tiny corrected-native-primitives CPU text fixture
Complete randomly initialized, untrained QFSDeepseekV4ForCausalLM text class
using Transformers5.16.1 native primitives and a reviewed RMSNorm arithmetic correction.
No upstream weights, paid GPU/cloud compute or useful-model claim.
This is not unmodified native Transformers or the complete production release.
Architecture and scope
Text lineage:… See the full description on the dataset page: https://huggingface.co/datasets/malaiwah/deepseek-v4-tiny-cpu-repro-v1.scale-swe-distill5000-deepseek-v4-flash-0731-think-rollout4-instance3393-trajectories7928
Scale-SWE DeepSeek V4 Flash 0731 Think Rollouts
Successful AweAgent trajectories generated with deepseek-v4-flash-0731 in think mode.
Dataset summary
Source task instances: 3,393
Rollouts per source instance: 4
Total attempted rollouts: 13,572
Successful exported trajectories: 7,928
Unique instances represented by successful trajectories: 2,250
Scaffold: aweagent
Tool-call format: openai_function
The export retains assistant reasoning_content, function tool… See the full description on the dataset page: https://huggingface.co/datasets/wjn922-01/scale-swe-distill5000-deepseek-v4-flash-0731-think-rollout4-instance3393-trajectories7928.DeepSeek-V4-Flash-0731-Teacher-Distillation-40513x
DeepSeek V4 Flash 0731 Teacher Distillation — 40,513 Retained Rows
Teacher-distillation corpus generated with
deepseek-ai/DeepSeek-V4-Flash-0731.
The original manifest contained 45,000 unique seeds.
Following generation, QC, retry-based repair, quarantine auditing,
and recovery adjudication, 40,513 rows were retained.
Composition
Bucket
Rows
Coding
5,601
Agentic
9,982
Cyber blue
13,000
Controlled cyber red
6,999
Tool use
4,931
Total
40,513… See the full description on the dataset page: https://huggingface.co/datasets/trjxter/DeepSeek-V4-Flash-0731-Teacher-Distillation-40513x.swebench-verified-deepseek-v4-flash-failure-analysis
SWE-bench Verified runs & failure analysis — DeepSeek-V4-flash (local) × mini-swe-agent
Per-instance analysis of SWE-bench Verified runs of a locally-served DeepSeek-V4-flash model
driven by mini-swe-agent, graded with the official
SWE-bench harness. Each instance carries the full agent trajectory, a readable transcript, the
submitted patch, the harness test output, deterministic metrics, and a hand-verified qualitative
root-cause diagnosis.
Current numbers (resolve rates… See the full description on the dataset page: https://huggingface.co/datasets/daaain/swebench-verified-deepseek-v4-flash-failure-analysis.Deepseek-V4-Reasoning-Code-2500
DeepSeek Reasoning and Code Distillation Dataset
This dataset contains synthetic instruction-response examples generated from coding, reasoning, and math prompts. It was generated with enforce_distillable_text enabled using DeepSeek V4 Pro and DeepSeek V4 Flash through OpenRouter. It is intended for experimentation with supervised fine-tuning, response-style distillation, reasoning-format analysis, and code-assistant behavior research.
The dataset file is:
train.csv
It contains 2… See the full description on the dataset page: https://huggingface.co/datasets/Banaxi-Tech/Deepseek-V4-Reasoning-Code-2500.Chinese-DeepSeek-R1-Distill-data-110k-decontaminated
Decontaminated — Congliu/Chinese-DeepSeek-R1-Distill-data-110k
What this is
A filtered version of Congliu/Chinese-DeepSeek-R1-Distill-data-110k (revision
8520b649430617c2be4490f424d251d09d835ed3) with exact-duplicate rows and rows overlapping standard benchmark test sets
removed. This is a different artifact from the companion contamination report — that one is an
audit of what's wrong; this one is the corpus with those rows actually taken out, ready to train on.… See the full description on the dataset page: https://huggingface.co/datasets/liodon-ai/Chinese-DeepSeek-R1-Distill-data-110k-decontaminated.deepseek-v4-pro-math-cot-1k
DeepSeek V4 Pro Math CoT 1K
A small, high-signal supervised-fine-tuning (SFT) dataset of math reasoning traces. Problems were sampled from a Nemotron math problem set (originally sourced from StackExchange-Math and AoPS), answered by DeepSeek V4 Pro with thinking enabled at high reasoning effort, then independently reviewed by DeepSeek V4 Flash for correctness against the expected answer. Pathological reasoning traces (looping, run-away length, excessive Wait-style backtracking)… See the full description on the dataset page: https://huggingface.co/datasets/blythet/deepseek-v4-pro-math-cot-1k.deepseek-v4-pro-0813-agentic
DeepSeek-V4-Pro 0813 Agentic (DS4)
A standalone, verifiable-first agentic training corpus: 19,072 training traces
plus 2,135 held-out evaluation rows (validation 1,070 / test 1,065), generated by
DeepSeek-V4-Pro 0813 (deepseek-v4-pro-0813, official API, thinking mode) across 13 verifiable task families,
each row admitted only after passing a deterministic programmatic verifier. The corpus is
designed to be directly usable for SFT, GRPO/RLVR, and NeMo Gym / NeMo RL
(verified… See the full description on the dataset page: https://huggingface.co/datasets/ArkhAngelLifeJiggy/deepseek-v4-pro-0813-agentic.deepseek-hermes-reasoning-traces
DeepSeek V4 Pro Hermes Reasoning Traces
19,331 multi-turn ChatML + Hermes reasoning traces generated by DeepSeek V4 Pro. Designed for LoRA fine-tuning local models to operate as Hermes Agent instances.
Quick Start
\
Splits
Split
Traces
train
16,431
valid
1,933
test
967
Variants (VRAM-Tiered)
Variant
Max Tokens
Traces
GPU
nano
2,048
15,948
Dev / 7B
budget
4,096
2,149
48GB
standard
8,192
990
64GB
spark
16,384
244… See the full description on the dataset page: https://huggingface.co/datasets/ansulev/deepseek-hermes-reasoning-traces.tau2-airline-deepseek-distill
τ²-bench airline · DeepSeek teacher trajectories
Successful multi-turn agent trajectories on τ²-bench's
airline domain, generated by running DeepSeek V4 Flash as the agent through the real τ²-bench
harness — same system prompt, same 14 tool schemas, same dialogue loop, same evaluator.
Used to behavior-clone the RL warm start
yuyu0529nya/qwen2.5-7b-tau2-airline-sft-lora,
which is the step 0 of the tau2_airline verl recipe.
Why these exist
GRPO on τ²-bench-airline… See the full description on the dataset page: https://huggingface.co/datasets/yuyu0529nya/tau2-airline-deepseek-distill.math500-cot-deepseek-r1-1.5b
MATH-500 CoT completions (DeepSeek-R1-Distill-Qwen-1.5B)
Successful chain-of-thought completions for HuggingFaceH4/MATH-500 test problems, generated with deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B via vLLM.
Files
File
Description
records.parquet
Main dataset: correct completions as token IDs
manifest.json
Schema, tokenizer, run ids, decoding config
problem_index.json
unique_id → problem_idx in MATH-500 test
subject_max_tokens.json
Per-subject completion… See the full description on the dataset page: https://huggingface.co/datasets/ChrisMcCormick/math500-cot-deepseek-r1-1.5b.llm_timeline_deepseek_v4_flash-pi
Coding agent session traces
This dataset contains coding agent session traces collected while working on LLM Timeline web app using the prompt from coding-agent-bench-prompts
swebench-deepseek-traces
swebench-deepseek-traces
Complete HTTP-level agentic traces from running swebench_deepseek benchmark tasks through an instrumented reverse proxy.
Each trace captures full request/response pairs including system prompts, user messages, assistant responses, tool calls and results, and token usage metadata.
Stats
Total sessions: 50
Multi-turn sessions (2+ LLM calls): 18
Total records: 542
Total LLM requests: 271
Format
Raw JSONL traces from the instrumented… See the full description on the dataset page: https://huggingface.co/datasets/sammshen/swebench-deepseek-traces.swebench-deepseek-traces-3
swebench-deepseek-traces-3
Complete HTTP-level agentic traces from running swebench_deepseek benchmark tasks through an instrumented reverse proxy.
Each trace captures full request/response pairs including system prompts, user messages, assistant responses, tool calls and results, and token usage metadata.
Stats
Total sessions: 46
Multi-turn sessions (2+ LLM calls): 26
Total records: 2154
Total LLM requests: 1077
Format
Raw JSONL traces from the instrumented… See the full description on the dataset page: https://huggingface.co/datasets/sammshen/swebench-deepseek-traces-3.deepseek-v4-flash-swe-cot
DeepSeek-V4-Flash SWE Agent Trajectories (with raw chain-of-thought)
795 multi-turn software-engineering agent trajectories generated by
DeepSeek-V4-Flash-0731 at reasoning_effort=max, each one executed in a real
repository inside an isolated container and verified by running the repository's own
tests. 469 are verified-correct.
Every assistant turn preserves reasoning_content — the model's raw chain-of-thought,
not a summary. That is the point of this dataset: the DeepSeek API… See the full description on the dataset page: https://huggingface.co/datasets/blythet/deepseek-v4-flash-swe-cot.Kimi-K3-And-DeepSeek-V4-Pro-0813-Distillation-in-PT-BR
🇧 Destilação PT-BR com Raciocínio (Chain-of-Thought)
Este dataset contém exemplos de alta qualidade gerados através da destilação de modelos de ponta (Teacher Models) disponíveis via NVIDIA NIM, focados em instrução, raciocínio lógico e naturalidade em Português Brasileiro (PT-BR).
O grande diferencial deste dataset é a inclusão explícita do processo de pensamento (Chain-of-Thought / thinking) dos modelos professores, permitindo treinar modelos menores (Student Models) não… See the full description on the dataset page: https://huggingface.co/datasets/Davizig10jojo/Kimi-K3-And-DeepSeek-V4-Pro-0813-Distillation-in-PT-BR.swebench-deepseek-traces-2
swebench-deepseek-traces-2
Complete HTTP-level agentic traces from running swebench_deepseek benchmark tasks through an instrumented reverse proxy.
Each trace captures full request/response pairs including system prompts, user messages, assistant responses, tool calls and results, and token usage metadata.
Stats
Total sessions: 37
Multi-turn sessions (2+ LLM calls): 33
Total records: 2040
Total LLM requests: 1020
Format
Raw JSONL traces from the instrumented… See the full description on the dataset page: https://huggingface.co/datasets/sammshen/swebench-deepseek-traces-2.denovoswe-distill2767-deepseek-v4-flash-0731-rollout8-instance674-trajectories3985
DeNovoSWE Distill 2767 — DeepSeek V4 Flash NL2Repo Trajectories
This dataset contains 3,985 difficulty-filtered NL2Repo SFT trajectories from 674 repository
instances. Each instance was sampled with eight rollouts using deepseek-v4-flash-0731.
Selection
Tasks receive a static difficulty score derived only from Stage 1–4 artifacts. The 1,141
successful tasks are split into five equal-count difficulty bins. A rollout is retained when its
evaluator score is greater… See the full description on the dataset page: https://huggingface.co/datasets/wjn922-01/denovoswe-distill2767-deepseek-v4-flash-0731-rollout8-instance674-trajectories3985.med-synth-questions-gemma-3-27b-deepseek-v4-flash
Med Synth Questions (Gemma-3 + DeepSeek V4 Flash)
Synthetic reasoning traces and answers for medical questions from openmed-community/med-synth-questions-gemma-3-27b-it. Each record contains a medical question with SYNTH-style reasoning and a generated answer by DeepSeek V4 Flash.
Dataset Summary
29,148 records (2 dupes + 3,410 incomplete/truncated removed from 32,560 source)
29,148 reasoning turns (99.2% format compliance)
Average 1,591 chars per reasoning trace… See the full description on the dataset page: https://huggingface.co/datasets/mkurman/med-synth-questions-gemma-3-27b-deepseek-v4-flash.deepseek-r1-systems-kernel-reasoning
🧠 DeepSeek-R1 Low-Level Systems & Kernel Reasoning Suite (2026)
🛒 Commercial Full Suite Available:
The full production suite with 10,000 SFT Hardware Reasoning Traces + 2,500 High-Contrast DPO Alignment Pairs across all 20 domains is available on Gumroad:
👉 Download Full Commercial Dataset on Gumroad (Starter \ / Pro \ / Enterprise )
A Tier-1 Commercial Dataset Suite engineered specifically for fine-tuning DeepSeek-R1, DeepSeek-R1-Distill-Qwen-14B/32B, and frontier… See the full description on the dataset page: https://huggingface.co/datasets/beatsprom/deepseek-r1-systems-kernel-reasoning.DeepSeek_0528_8B_Legal_Distill
⚖️ DeepSeek-0528-8B Legal Distill Dataset
This repository contains a high-density trajectory dataset generated during the GRPO (Group Relative Policy Optimization) training of the DeepSeek-8B architecture. It is specifically optimized for advanced Knowledge Distillation and structural legal reasoning. 🚀
💡 The Concept: "Log-as-Distillation"
Traditional training often treats logs as temporary metadata. This dataset flips that script. By capturing the multi-generation… See the full description on the dataset page: https://huggingface.co/datasets/Azzindani/DeepSeek_0528_8B_Legal_Distill.deepseek-v4-pro-0813-agentic
DeepSeek-V4-Pro 0813 Agentic (DS4)
A standalone, verifiable-first agentic training corpus: 19,072 training traces
plus 2,135 held-out evaluation rows (validation 1,070 / test 1,065), generated by
DeepSeek-V4-Pro 0813 (deepseek-v4-pro-0813, official API, thinking mode) across 13 verifiable task families,
each row admitted only after passing a deterministic programmatic verifier. The corpus is
designed to be directly usable for SFT, GRPO/RLVR, and NeMo Gym / NeMo RL
(verified… See the full description on the dataset page: https://huggingface.co/datasets/ansulev/deepseek-v4-pro-0813-agentic.