datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Qwen3.8-27B-GGUF-metrics
Qwen3.8-27B GGUF, everything behind the numbers
This is the working record for
AtomicChat/Qwen3.8-27B-GGUF.
Every figure in that model card came from a file in here, including the ones
about other publishers' builds.
The point of publishing it is simple. A quantization comparison is only worth
reading if someone else can run it, and that needs three things nobody usually
ships: the exact reference the numbers were measured against, the exact text
they were measured on, and the… See the full description on the dataset page: https://huggingface.co/datasets/AtomicChat/Qwen3.8-27B-GGUF-metrics.qwen3.8-max-glm5.2-kimi-k3-distillation
Multi-Teacher Distillation Dataset (57,937 traces)
A quality-filtered, deduplicated, multi-teacher SFT corpus combining traces from three frontier models across math, code, reasoning, instruction-following, tool-use, science, long-context, multilingual, and creative dialogue domains.
Teachers
Teacher
Provider
Traces
Qwen3.8-Max-Preview
Alibaba Cloud Model Studio
48,283
GLM-5.2
Z.AI Coding Plan
5,307
Kimi Code K3
Moonshot AI (Kimi)
4,347… See the full description on the dataset page: https://huggingface.co/datasets/r0b0tlab/qwen3.8-max-glm5.2-kimi-k3-distillation.MMFineReason-Full-2.3M-Qwen3-VL-235B-Thinking
MMFineReason-Full-2.3M
The Complete Pre-Selection Dataset — Before Quality Filtering
📖 Overview
MMFineReason-Full-2.3M is the complete pre-selection dataset containing 2.3M samples and 8.8B solution tokens, generated through our reasoning distillation pipeline before the data selection stage. This dataset includes all samples that passed basic template and length validation, but have not undergone correctness verification filtering.
🎯 Key Characteristics… See the full description on the dataset page: https://huggingface.co/datasets/OpenDataArena/MMFineReason-Full-2.3M-Qwen3-VL-235B-Thinking.Qwen3.8-27B-Distillation-40K
Qwen3.8-27B-Distillation (40K Traces)
Qwen3.8-27B-Distillation is a dataset containing 40,000 reasoning traces distilled from Qwen's latest model — Qwen3.8-27B. We generated this dataset locally by running the model on our own infrastructure. It covers 4 domains with prompts sourced from 12 diverse open-source datasets.
Dataset Overview
Metric
Value
Total Examples
40,000
Teacher Model
Qwen3.8-27B
Model Precision
FP8
Reasoning Effort
medium… See the full description on the dataset page: https://huggingface.co/datasets/faunix/Qwen3.8-27B-Distillation-40K.context_qa_sum_qwen3_synthetic
Context-based QA and Summarization Synthetic Dataset
Overview
This dataset contains synthetic context-based question-answering (QA) and summarization data. The data was synthesized using:
Source context: openbmb/Ultra-FineWeb
Synthesis model: Qwen3-30B-A3B-Instruct-2507
Each context is obtained by taking the initial segment of raw pretraining text from Ultra-FineWeb, truncated to at most the corresponding number of tokens, while ensuring the truncation does not occur in… See the full description on the dataset page: https://huggingface.co/datasets/yuyijiong/context_qa_sum_qwen3_synthetic.qwen36-27b-length-traces
Qwen3.6-27B generation-length prediction: heads, calibrations and workloads
Artifacts for conformal length-aware LLM scheduling on Qwen/Qwen3.6-27B — predicting a
request's remaining generation length from a hidden layer during decoding, wrapping it in a
split-conformal interval, and scheduling with SRPT inside vLLM. Extends TRAIL
(Don't Stop Me Now, ICLR'25) to a hybrid-attention reasoning model.
This repo contains the derived artifacts, not the raw activations. The 3250… See the full description on the dataset page: https://huggingface.co/datasets/dungnv/qwen36-27b-length-traces.MMFineReason-1.8M-Qwen3-VL-235B-Thinking
MMFineReason
Closing the Multimodal Reasoning Gap via Open Data-Centric Methods
Average score across mathematical reasoning and multimodal understanding benchmarks.
📖 Overview
MMFineReason is a large-scale, high-quality multimodal reasoning dataset comprising 1.8M samples and 5.1B solution tokens, featuring detailed reasoning annotations distilled from Qwen3-VL-235B-A22B-Thinking.
🎯 Key Highlights
1.8M High-Quality Samples with 5.1B Solution Tokens… See the full description on the dataset page: https://huggingface.co/datasets/OpenDataArena/MMFineReason-1.8M-Qwen3-VL-235B-Thinking.MMFineReason-Full-2.3M-Qwen3-VL-235B-Thinking
MMFineReason-Full-2.3M
The Complete Pre-Selection Dataset — Before Quality Filtering
📖 Overview
MMFineReason-Full-2.3M is the complete pre-selection dataset containing 2.3M samples and 8.8B solution tokens, generated through our reasoning distillation pipeline before the data selection stage. This dataset includes all samples that passed basic template and length validation, but have not undergone correctness verification filtering.
🎯 Key Characteristics… See the full description on the dataset page: https://huggingface.co/datasets/ericktwo/MMFineReason-Full-2.3M-Qwen3-VL-235B-Thinking.Math-CoT-44k-Qwen3-32b-n32-16384-with-logprob-and-entropy
Qwen3-32B Math n32 16384 (44k Queries)
This dataset contains multi-sampled rollout traces from Qwen3-32B on around 44k math queries.
For each query, the model is rolled out 32 times with a maximum generation length of 16384 tokens.
Each response is annotated with answer correctness (acc_reward), and includes token-level statistics (action_entropy, action_log_probs) for further analysis and research.
Resources
Paper: Rethinking Generalization in Reasoning SFT: A… See the full description on the dataset page: https://huggingface.co/datasets/jasonrqh/Math-CoT-44k-Qwen3-32b-n32-16384-with-logprob-and-entropy.FineReason-1.8M-Qwen3-VL-235B-Thinking
MMFineReason
Closing the Multimodal Reasoning Gap via Open Data-Centric Methods
Average score across mathematical reasoning and multimodal understanding benchmarks.
📖 Overview
MMFineReason is a large-scale, high-quality multimodal reasoning dataset comprising 1.8M samples and 5.1B solution tokens, featuring detailed reasoning annotations distilled from Qwen3-VL-235B-A22B-Thinking.
🎯 Key Highlights
1.8M High-Quality Samples with 5.1B Solution Tokens… See the full description on the dataset page: https://huggingface.co/datasets/NarsAI/FineReason-1.8M-Qwen3-VL-235B-Thinking.Qwen3.6-35B-A3B-mcr-stage-b
Qwen3.6-35B-A3B — MCR Stage B Corpus (Distributed Reasoning Localization)
First systematic mechanistic-intervention corpus on a hybrid MoE + GDN + Gated-Attention architecture.
📄 Paper: Loop-Intolerance Profiling: Localizing Distributed Reasoning in a Hybrid MoE Architecture via Nine Convergent Intervention Experiments — submitted to arXiv (2026-04-20, in moderation). Final arXiv ID will be added here once approved.
This dataset contains per-token residual-stream activations at… See the full description on the dataset page: https://huggingface.co/datasets/caiovicentino1/Qwen3.6-35B-A3B-mcr-stage-b.qwen3.8-max-glm5.2-kimi-k3-distillation
Multi-Teacher Distillation Dataset (57,937 traces)
A quality-filtered, deduplicated, multi-teacher SFT corpus combining traces from three frontier models across math, code, reasoning, instruction-following, tool-use, science, long-context, multilingual, and creative dialogue domains.
Teachers
Teacher
Provider
Traces
Qwen3.8-Max-Preview
Alibaba Cloud Model Studio
48,283
GLM-5.2
Z.AI Coding Plan
5,307
Kimi Code K3
Moonshot AI (Kimi)
4,347… See the full description on the dataset page: https://huggingface.co/datasets/o0Biggz0o/qwen3.8-max-glm5.2-kimi-k3-distillation.slimder-qwen38-ngram-activation-refine-20260901
REAM-288 activation-aware 50% PLE selection
This dataset contains the verified capacity-preserving row selection and remap
tables for compacting the PLE n-gram memory in
sjakek/slimder-qwen38-ream288-depth32-agentic to exactly 50% of its padded
physical capacity.
The selection combines a 250M-token multi-domain frequency census with a
3M-token proportional-stratified PLE activation-saliency pass. The saliency
sample contains 900k code, 750k agentic, 450k retrieval, 300k… See the full description on the dataset page: https://huggingface.co/datasets/jakeatx/slimder-qwen38-ngram-activation-refine-20260901.Qwen3.8-27B-Drafter-SFT
Qwen3.8-27B Drafter SFT Corpus
Supervised fine-tuning data released for training speculative drafters for Qwen/Qwen3.8-27B. All completions were generated with Qwen/Qwen3.8-27B at revision 1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0.
The dataset contains 367,535 source conversations and 450,401 train rows, totaling 1,953,218,671 tokens after filtering and evaluation decontamination. Rows contain Qwen3.8-27B-tokenized prompts and target-generated completions, together with loss… See the full description on the dataset page: https://huggingface.co/datasets/DaoCloud/Qwen3.8-27B-Drafter-SFT.qwen38-27b-fidelity-suite-v3
Qwen3.8-27B distribution-fidelity suite v3 (held-out, 181 x 2048)
The frozen evaluation suite and the captured hidden states that let anyone recompute or
contest the KL-divergence numbers published for
malaiwah/Qwen3.8-27B-K4 — without a
GPU, without downloading any model, and without trusting the publisher.
Scope, stated up front. The captures in this snapshot are the iteration-1 set:
malaiwah/Qwen3.8-27B-K4, unsloth/Qwen3.8-27B-NVFP4 and Qwen/Qwen3.8-27B-FP8 against
the BF16… See the full description on the dataset page: https://huggingface.co/datasets/malaiwah/qwen38-27b-fidelity-suite-v3.MMFineReason-1.8M-Qwen3-VL-235B-Thinking
MMFineReason
Closing the Multimodal Reasoning Gap via Open Data-Centric Methods
Average score across mathematical reasoning and multimodal understanding benchmarks.
📖 Overview
MMFineReason is a large-scale, high-quality multimodal reasoning dataset comprising 1.8M samples and 5.1B solution tokens, featuring detailed reasoning annotations distilled from Qwen3-VL-235B-A22B-Thinking.
🎯 Key Highlights
1.8M High-Quality Samples with 5.1B Solution Tokens… See the full description on the dataset page: https://huggingface.co/datasets/Sandeepthakur/MMFineReason-1.8M-Qwen3-VL-235B-Thinking.qwen3.8-max-distillation-50k
Qwen3.8-Max Distillation 50K
A curated dataset of 49,772 teacher-generated traces from qwen3.8-max-preview, prepared for supervised fine-tuning and off-policy knowledge distillation.
The teacher responses are preserved as returned by the API. Where the model emitted visible <think>...</think> blocks, those blocks remain in the assistant message. Some simpler prompts received direct answers without a thinking block.
[!CAUTION]
Terms and provenance notice — not cleared for… See the full description on the dataset page: https://huggingface.co/datasets/r0b0tlab/qwen3.8-max-distillation-50k.qwen3.7-max-pi-tracesThis 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.
Qwen3.7 Max Pi Traces
This directory contains raw agent trace files generated by teich.
All assistant responses were generated by qwen/qwen3.7-max.
JSONL files: 47
Training-ready tools
A complete configured tools schema snapshot is embedded in the collapsed section at the bottom of this README.
Use it… See the full description on the dataset page: https://huggingface.co/datasets/armand0e/qwen3.7-max-pi-traces.qwen3.8-max-glm5.2-kimi-k3-distill
Multi-Teacher Distillation Dataset (57,937 traces)
A quality-filtered, deduplicated, multi-teacher SFT corpus combining traces from three frontier models across math, code, reasoning, instruction-following, tool-use, science, long-context, multilingual, and creative dialogue domains.
Teachers
Teacher
Provider
Traces
Qwen3.8-Max-Preview
Alibaba Cloud Model Studio
48,283
GLM-5.2
Z.AI Coding Plan
5,307
Kimi Code K3
Moonshot AI (Kimi)
4,347… See the full description on the dataset page: https://huggingface.co/datasets/ansulev/qwen3.8-max-glm5.2-kimi-k3-distill.qwen3.8-max-glm5.2-kimi-k3-distillation
Multi-Teacher Distillation Dataset (57,937 traces)
A quality-filtered, deduplicated, multi-teacher SFT corpus combining traces from three frontier models across math, code, reasoning, instruction-following, tool-use, science, long-context, multilingual, and creative dialogue domains.
Teachers
Teacher
Provider
Traces
Qwen3.8-Max-Preview
Alibaba Cloud Model Studio
48,283
GLM-5.2
Z.AI Coding Plan
5,307
Kimi Code K3
Moonshot AI (Kimi)
4,347… See the full description on the dataset page: https://huggingface.co/datasets/p-research/qwen3.8-max-glm5.2-kimi-k3-distillation.qwen3.8-max-glm5.2-kimi-k3-distillation
Multi-Teacher Distillation Dataset (57,937 traces)
A quality-filtered, deduplicated, multi-teacher SFT corpus combining traces from three frontier models across math, code, reasoning, instruction-following, tool-use, science, long-context, multilingual, and creative dialogue domains.
Teachers
Teacher
Provider
Traces
Qwen3.8-Max-Preview
Alibaba Cloud Model Studio
48,283
GLM-5.2
Z.AI Coding Plan
5,307
Kimi Code K3
Moonshot AI (Kimi)
4,347… See the full description on the dataset page: https://huggingface.co/datasets/inferenceport-ai/qwen3.8-max-glm5.2-kimi-k3-distillation.32B-reason-only.stride-1.k-8.statml-arxiv.qwen3-ids
32B-reason-only.stride-1.k-8.statml-arxiv.qwen3-ids
Thoughts for next-token prediction on k=8 token chunks of JackHsieh/statML-arxiv-40M-20M, generated by Qwen3-32B
with thinking mode off. Each thought is a few dense sentences of reasoning about the next
8 tokens after a cut, written from the document prefix alone — the generator never sees the
continuation. Stored thought_text includes the <thought>/</thought> wrapper.
The 4B parity counterpart is… See the full description on the dataset page: https://huggingface.co/datasets/JackHsieh/32B-reason-only.stride-1.k-8.statml-arxiv.qwen3-ids.maxrl_qwen3_4B_base_polaris_rollouts
MaxRL Qwen3-4B-Base training rollouts (POLARIS math prompts)
Every training rollout from an online RL run, with exact token ids, sampling
log-probs, and raw rewards — usable as a replay buffer to study off-policy RL
for LLM reasoning completely offline.
The run: Qwen3-4B-Base trained with the maxRL advantage estimator
(A = (r - mean)/(mean + eps), group mean over 16 rollouts per prompt;
maxRL paper) and a pure REINFORCE loss
(L = -A * log pi; no importance ratio, no clipping, no… See the full description on the dataset page: https://huggingface.co/datasets/ftajwar/maxrl_qwen3_4B_base_polaris_rollouts.qwen3.8-flash-next-expert-traces
Qwen3.8-Flash-Next expert routing traces
Token-level routing traces of a deployed MoE model: for every token and every one of the
48 MoE layers, which experts the router chose, the top-32 router logits behind that choice,
and the exact hidden state the router read — plus, in v3, the state at many layers per token,
the post-final-norm state the LM head consumes, and the LM head's top-8 next-token candidates.
The corpus exists to answer one question: how well can the next tokens'… See the full description on the dataset page: https://huggingface.co/datasets/aswinkumar99/qwen3.8-flash-next-expert-traces.4B-Instruct-reason-only.stride-1.k-8.statml-arxiv.qwen3-ids
4B-Instruct-reason-only.stride-1.k-8.statml-arxiv.qwen3-ids
Thoughts for next-token prediction on k=8 token chunks of JackHsieh/statML-arxiv-40M-20M, generated by
Qwen3-4B-Instruct-2507. Each thought is a few dense sentences of reasoning about the next
8 tokens after a cut, written from the document prefix alone — the generator never sees the
continuation. Stored thought_text includes the <thought>/</thought> wrapper.
This is the small-generator parity counterpart of… See the full description on the dataset page: https://huggingface.co/datasets/JackHsieh/4B-Instruct-reason-only.stride-1.k-8.statml-arxiv.qwen3-ids.qwen3.8-max-glm5.2-distillation-51389
Qwen3.8-Max / GLM-5.2 Distillation — 51,389 Rows
A deterministic, public Parquet release of admitted teacher traces for supervised fine-tuning, reasoning-format studies, tool-use studies, and tokenizer-specific rendering experiments. The sft configuration is the default training view. The package contains data and documentation only; it does not require executable dataset code.
Credits and Attribution
Dataset assembly and release packaging: r0b0tlab.
Qwen-derived… See the full description on the dataset page: https://huggingface.co/datasets/ufrik/qwen3.8-max-glm5.2-distillation-51389.MMFineReason-SFT-123K-Qwen3-VL-235B-Thinking
MMFineReason-SFT-123K
The Hardest 7% — Less Data, More Reasoning
📖 Overview
MMFineReason-SFT-123K is a difficulty-filtered subset of MMFineReason-1.8M, containing only the hardest 7% of samples where Qwen3-VL-4B-Thinking consistently fails (pass rate = 0).
🎯 Key Highlights
123K Challenging Samples: Only instances where a 4B thinking model fails all 4 inference attemptsEfficient Training: Comparable performance to full 1.8M dataset with only 7% of… See the full description on the dataset page: https://huggingface.co/datasets/OpenDataArena/MMFineReason-SFT-123K-Qwen3-VL-235B-Thinking.qwen38-27b-fidelity-suite-v5
Qwen3.8-27B fidelity suite v5 — evaluation inputs, the BF16 reference, and every per-shard report
This dataset exists because we deleted expensive artifacts once and had to remake them. Every
tree here is replayable input for a future candidate, not a finished result — the finished
results live as receipts in the research repo. Publishing the inputs means the next candidate
costs one download instead of a fresh BF16 capture, and it means anyone can check our numbers
without our… See the full description on the dataset page: https://huggingface.co/datasets/malaiwah/qwen38-27b-fidelity-suite-v5.AM-Qwen3-Distilled
📘 Dataset Summary
AM-Thinking-v1 and Qwen3-235B-A22B are two reasoning datasets distilled from state-of-the-art teacher models. Each dataset contains high-quality, automatically verified responses generated from a shared set of 1.89 million queries spanning a wide range of reasoning domains.
The datasets share the same format and verification pipeline, allowing for direct comparison and seamless integration into downstream tasks. They are intended to support the development of… See the full description on the dataset page: https://huggingface.co/datasets/a-m-team/AM-Qwen3-Distilled.qwen3.8-max-glm5.2-kimi-k3-distillation
Multi-Teacher Distillation Dataset (57,937 traces)
A quality-filtered, deduplicated, multi-teacher SFT corpus combining traces from three frontier models across math, code, reasoning, instruction-following, tool-use, science, long-context, multilingual, and creative dialogue domains.
Teachers
Teacher
Provider
Traces
Qwen3.8-Max-Preview
Alibaba Cloud Model Studio
48,283
GLM-5.2
Z.AI Coding Plan
5,307
Kimi Code K3
Moonshot AI (Kimi)
4,347… See the full description on the dataset page: https://huggingface.co/datasets/bhadra123/qwen3.8-max-glm5.2-kimi-k3-distillation.