Get started Console

Modern medicine75

Traditional medicine24

Cross-cutting12

Collections

Models8

License status and Mdexa hosting, recorded separatelyAll datasets

DISC-Med-SFT

Verified by Mdexa · Checked against the source text under public rules.How we check

Commercial use
No
Derivatives
UnclearUnstated
Redistribution
UnclearUnstated
File on Mdexa
NoLink only

Declares Apache-2.0, but upstream cMedQA2 is limited to non-commercial research and a MedDialog mirror says copyright belongs to haodf.com

RestrictedLink onlyNOASSERTIONSave

Badge

Mdexa verified · Restricted · 2026-11Mdexa verifiedRestricted · 2026-11

[![Mdexa verified · Restricted · 2026-11](https://mcp.mdexa.com/v1/badge/datasets/disc-med-sft.svg)](https://mdexa.com/datasets/disc-med-sft/)

Place it on a repository or dataset page; the status and date follow the verification. About badges

Cite this pageMdexa, DISC-Med-SFT, https://mdexa.com/datasets/disc-med-sft/, verified 2026-11-12.

Source riskDerivedScrapedReal patient textModel-generated

Key facts

Domain
Clinical medicine Modern medicine
Content
QA, dialogue and instruction dataConsultation and knowledge QA fine-tuning data
Size
  • ≈470,000 itemsGitHub
  • 464,898 rowsHF subset
As stated: GitHub 称近47万条；HF 子集 464,898行
Language / modality
Chinese / Text
Publisher
复旦大学 DISC 实验室（FudanDISC）
Year
2023 (arXiv 2308.14346)
Expert review
2k human-curated preference items; not stated for the rest
Verification
Verified: license text seen

Commercial use of DISC-Med-SFT

Commercial use: No. License status: Restricted.

Quoted statements and their sources are in DISC-Med-SFT license evidence.

Verdicts record what sources state and are not legal advice.

Tags

Task format
Multi-turn dialogue

Quoted from the source* **多轮对话的问询能力**，我们以真实咨询对话纪录作为信息源，使用大模型进行对话重建，构建过程中要求模型完全对齐对话中的医学信息。Snapshot s03

Open-ended QA
Intended use
Fine-tuning

Quoted from the source我们总共发布了近47万条训练数据，其中包括重构AI医患对话和知识图谱问答对。您可以访问这个[链接](https://huggingface.co/datasets/Flmc/DISC-Med-SFT)下载数据集。Snapshot s03

Specialty
Multi-specialty

Author-statedInferredHuman-confirmed

DISC-Med-SFT license evidence 4

The strictest upstream applies: restricted

Code repository license: Apache-2.0. This licenses the repository, not the data. LICENSE file · verified s04

  1. Hugging Face card metadataPage opened
    Apache-2.0
    Third-party text · Hugging Face card metadatalicense: apache-2.0

    HF dataset card: Apache-2.0

    huggingface.co2026-11-12 · s01
  2. Repository license filePage opened
    Apache-2.0
    Third-party text · Repository license fileApache License Version 2.0, January 2004

    GitHub repository Apache-2.0; no separate data license

    github.com2026-11-12 · s04
  3. Upstream termsPage opened
    Third-party text · Upstream termsavailable for non-commercial research

    Upstream cMedQA2 README (cmedqa2 card, snapshot s01): non-commercial research only

    github.com2026-11-12
  4. Upstream termsSecond-hand only
    Third-party text · Upstream termsAll copyrights of the data belong to haodf.com.

    Upstream MedDialog third-party mirror (meddialog-cn card); secondhand

    Second-hand, not verified

    huggingface.co2026-11-12

Source and lineage

Restructured from existing datasets: MedDialog 400k, cMedQA2 20k, CMeKG 50k, human preference data 2k, MedMCQA 8k, MOSS-SFT 33k, Alpaca-GPT4-zh 1k

  • IncludesMedDialog
  • IncludescMedQA2
  • IncludesCMeKGnot in the database
  • IncludesMedMCQAnot in the database
  • IncludesMOSS-SFTnot in the database
  • IncludesAlpaca-GPT4-zhnot in the database

Known issues with DISC-Med-SFT

  • Upstream includes MedDialog (copyright belongs to haodf.com) and cMedQA2 (non-commercial research statement), inconsistent with the Apache-2.0 label
  • HF has only a subset (without the behavior preference data)
  • Part of the data was reconstructed or generated by an LLM (snapshot s03, quoted in the zh text).

Preflight

Not run Runs only on ingested data; says nothing about quality.

Downloading DISC-Med-SFT and API access

No file on Mdexa (see the verdict). Get it from the source and confirm the terms there:

The terms are restricted; no download command is given. See the verdict.

Citing DISC-Med-SFT

Mdexa, DISC-Med-SFT, https://mdexa.com/datasets/disc-med-sft/, verified 2026-11-12.
More citation details

The source gives no citation text; the line above cites this card.

Machine-readable

disc-med-sft.jsondisc-med-sft.mdcatalog.json

Change log 9
2026-11-13License status changed from conflict to restricted and hosting from pending to link-only: the declared license is broader than its upstream sources, so the strictest upstream applies.
2026-11-13Added specialty tags: multi-specialty(inferred).
2026-11-13Added the model-generated risk flag (the dialogues were rebuilt with a language model). Status and hosting unchanged.
2026-11-13Removed the stored size-tier tag; the tier is now computed from the size field when pages are built.
2026-11-13Added dataset-level tags: 4 (2 stated by the authors with excerpts, 2 inferred from the content). No conclusion changed.
2026-11-12Recorded the code repository license separately; it is not treated as a data license statement.
2026-11-12Added verbatim license excerpts and snapshot ids from captured page snapshots; no conclusion changed.
2026-11-12License status split into license_status (what the license says) and hosting (what Mdexa does); the earlier conclusion is kept for traceability.
2026-11-12Card converted from an earlier manual catalog; no new facts added; reclassified under taxonomy v2.

Datasets similar to DISC-Med-SFT

  • CMtMedQAVerified by MdexaMulti-turn dialogueFine-tuning PermissiveLink only ≈68K dialogues

    License is permissive, but source risk: derived from real patient text; link only.

    Clinical medicine · QA, dialogue and instruction data · Real patient text

    Open the cardSave

    Shared: Intended use Fine-tuning · Task format Multi-turn dialogue · Specialty Multi-specialty · domain Clinical medicine · content QA, dialogue and instruction data

  • Huatuo-Llama-Med-Chinese 知识库问答与文献数据Verified by MdexaMulti-turn dialogueFine-tuning RestrictedLink only ≥8,000 QA pairs

    License is restricted: README prohibits commercial use; link only.

    Clinical medicine · Biomedical literature · QA, dialogue and instruction data · Model-generated

    Open the cardSave

    Shared: Intended use Fine-tuning · Task format Multi-turn dialogue · Specialty Multi-specialty · domain Clinical medicine · content QA, dialogue and instruction data

  • BianQueCorpus扁鹊健康大数据Verified by MdexaMulti-turn dialogueFine-tuning RestrictedLink only 10M items (claimed)

    License is restricted: The official README limits the project to non-commercial research; under "restrictions are read broadly" this applies to the corpus; link only.

    Clinical medicine · QA, dialogue and instruction data · Derived · Scraped · Real patient text

    Open the cardSave

    Shared: Intended use Fine-tuning · Task format Multi-turn dialogue · Specialty Multi-specialty · domain Clinical medicine · content QA, dialogue and instruction data

  • ChatMed_Consult_DatasetVerified by MdexaOpen-ended QAFine-tuning ConflictPending 549.3K rows

    License conflict: HF metadata CC BY 4.0 conflicts with card text CC BY-NC 4.0; link only until resolved.

    Clinical medicine · QA, dialogue and instruction data · Scraped · Model-generated

    Open the cardSave

    Shared: Intended use Fine-tuning · Task format Open-ended QA · Specialty Multi-specialty · domain Clinical medicine · content QA, dialogue and instruction data

  • Chinese-medical-dialogue-dataToyhomVerified by MdexaOpen-ended QAFine-tuning PermissiveLink only 792.1K QA pairs

    License is permissive, but source risk: source not stated, so source copyright cannot be judged; link only.

    Clinical medicine · QA, dialogue and instruction data · Unknown provenance

    Open the cardSave

    Shared: Intended use Fine-tuning · Task format Open-ended QA · Specialty Multi-specialty · domain Clinical medicine · content QA, dialogue and instruction data