Papers › MEDITRON-70B: Scaling Medical Pretraining for Large Language Models

MEDITRON-70B: Scaling Medical Pretraining for Large Language Models

27 Nov 2023arXiv:2311.16079archive 2025-07-28

Zeming Chen, Alejandro Hernández Cano, Angelika Romanou, Antoine Bonnet, Kyle Matoba, Francesco Salvi, Matteo Pagliardini, Simin Fan, Andreas Köpf, Amirkeivan Mohtashami, Alexandre Sallinen, Alireza Sakhaeirad, Vinitra Swamy, Igor Krawczuk, Deniz Bayazit, Axel Marmet, Syrielle Montariol, Mary-Anne Hartley, Martin Jaggi, Antoine Bosselut

Large language models (LLMs) can potentially democratize access to medical knowledge. While many efforts have been made to harness and improve LLMs' medical knowledge and reasoning capacities, the resulting models are either closed-source (e.g., PaLM, GPT-4) or limited in scale (<= 13B parameters), which restricts their abilities. In this work, we improve access to large-scale medical LLMs by releasing MEDITRON: a suite of open-source LLMs with 7B and 70B parameters adapted to the medical domain. MEDITRON builds on Llama-2 (through our adaptation of Nvidia's Megatron-LM distributed trainer), and extends pretraining on a comprehensively curated medical corpus, including selected PubMed articles, abstracts, and internationally-recognized medical guidelines. Evaluations using four major medical benchmarks show significant performance gains over several state-of-the-art baselines before and after task-specific finetuning. Overall, MEDITRON achieves a 6% absolute performance gain over the best public baseline in its parameter class and 3% over the strongest baseline we finetuned from Llama-2. Compared to closed-source LLMs, MEDITRON-70B outperforms GPT-3.5 and Med-PaLM and is within 5% of GPT-4 and 10% of Med-PaLM-2. We release our code for curating the medical pretraining corpus and the MEDITRON model weights to drive open-source development of more capable medical LLMs.

PaperPDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 14 code samples harvested from 1 repository linked to this paper: 9 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 0 honoured, 0 violated, 9 with no contract checked; 0 where Syntology's instrument failed) · 5 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code; here there are none.

By repository: official repository: 14 samples from 1 repository, 9 ran (9 with no instrument failure; 0 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written on 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

epfllm/meditron officialmentioned in papermentioned on GitHubpytorchApache-2.0Syntology: 14 samples, 9 ran (0 instrument-failed) · 5 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

14 samples harvested: 9 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 0 honoured, 0 violated, 9 with no contract checked; 0 where Syntology's instrument failed) · 5 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

9ran
5unverified

Licence: 0 of the 14 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from epfllm/meditron. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity.

epfllm/meditron official repository 14 samples, 9 ran (0 instrument-failed) · 5 unverified
clean_mcq_answer evaluation/evaluate.py ran fingerprinted Apache-2.0 (permissive) · 4452a22b84309806 · report fixture: signature output:str/0/str · returned str · level recorded 2026-09-24
concatenate_sections gap-replay/guidelines/clean.py ran Apache-2.0 (permissive) · aeb12ea36b675148 · report fixture: signature article:dict/0/? · returned str · level recorded 2026-09-24
format_mcq evaluation/benchmarks.py ran fingerprinted Apache-2.0 (permissive) · 983fd1e0e6e34a3f · report fixture: signature question:str/0/str, options:list/1/str · returned str · level recorded 2026-09-24
format_prompt in 45 papers evaluation/inference.py ran Apache-2.0 (permissive) · 1ababc7e1e681f41 · report fixture: signature prompt:str/0/str, args:Args/0/? · returned str · level recorded 2026-09-24
get_mesh_tags gap-replay/pubmed/augment.py ran Apache-2.0 (permissive) · 15943fc5c02ac67e · report fixture: signature article:Element/0/? · returned tuple · level recorded 2026-09-24
infer_ndocs finetuning/sft.py ran Apache-2.0 (permissive) · 8143fb4298119175 · report fixture: signature cmd:list/1/str, autoaccept_iters:bool/0/bool · returned int · level recorded 2026-09-24
load_json in 344 papers evaluation/evaluate.py ran Apache-2.0 (permissive) · 7e8c378eedc74237 · report fixture: signature filename:str/0/str · returned dict · level recorded 2026-09-24
load_jsonl in 179 papers evaluation/evaluate.py ran Apache-2.0 (permissive) · ed2919d070e3e6e8 · report fixture: signature filename:str/0/str · returned list · level recorded 2026-09-24
read_jsonl in 136 papers gap-replay/guidelines/clean.py ran Apache-2.0 (permissive) · 63ecbb0137ec2cf0 · report fixture: signature path:str/0/str · returned list · level recorded 2026-09-24
benchmark_factory evaluation/benchmarks.py unverified Apache-2.0 (permissive) · df54cf786ee59bad · report
load_instruction evaluation/benchmarks.py unverified Apache-2.0 (permissive) · baf77bfa09d1505d · report
tokenize_data in 8 papers finetuning/sft.py unverified Apache-2.0 (permissive) · 42f0d9df1f4e7733 · report
tokenizer_param evaluation/inference.py unverified Apache-2.0 (permissive) · bc10fd06d25d060b · report
update_metadata_mesh gap-replay/pubmed/augment.py unverified Apache-2.0 (permissive) · 6c12a5dfd0746a7c · report

Tasks

ArticlesConditional Text GenerationFew-Shot LearningMultiple Choice Question Answering (MCQA)Question AnsweringZero-Shot Learning

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Few-Shot Learning MedConceptsQA epfl-llm/meditron-70b Accuracy 25.262 #7 of 12 Archive leaderboard report
Few-Shot Learning MedConceptsQA epfl-llm/meditron-7b Accuracy 23.787 #12 of 12 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MedMCQA Meditron-70B (CoT + SC) Dev Set (Acc-%) 66.0 #11 of 22 Archive leaderboard report
Question Answering MedQA Meditron-70B (CoT + SC) Accuracy 70.2 #9 of 27 Archive leaderboard report
Question Answering MedQA LLAMA-2 (70B SC CoT) Accuracy 61.5 #11 of 27 Archive leaderboard report
Question Answering MedQA LLAMA-2 (70B) Accuracy 59.2 #14 of 27 Archive leaderboard report
Question Answering PubMedQA Meditron-70B (CoT + SC) Accuracy 81.6 #1 of 30 Archive leaderboard report
Zero-Shot Learning MedConceptsQA epfl-llm/meditron-7b Accuracy 25.751 #5 of 13 Archive leaderboard report
Zero-Shot Learning MedConceptsQA epfl-llm/meditron-70b Accuracy 25.360 #8 of 13 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

AttentionBPECosine AnnealingDense ConnectionsDropoutGPT-4LLaMALabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPaLMPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformer

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections