SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.04449 · 2024

MAIRA-2: Grounded Radiology Report Generation

arXiv · PDF · Open in the Atlas

Code that ran

4 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 4 failures of our instrument; 0 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2406.04449.json · badge: /api/ran/2406.04449.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
microsoft/RadFact canonical (classifier label)
classifier basis: census
4 ran · 0 no recorded run
FunctionStatusWhere it lives
find_best_match Ran
ran; our contract draft was wrong, not the code
microsoft/RadFact/src/radfact/llm_utils/text_utils.py
code served (permissive licence) · get_code("3e69a2e27036eafd")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import logging
import re
from difflib import SequenceMatcher

logger = logging.getLogger(__name__)

def normalise_text_for_comparison(text: str) -> str:
    """
    Normalise a string for comparison by removing whitespace, dashes, and punctuation.

    :param text: The text to normalise.
    :return: The normalised text.
    """
    text = text.lower().strip()
    text = re.sub('\\s+', ' ', text)
    text = re.sub('[\\-–—]', '', text)
    text = re.sub('\\s*(?<=[\\.\\:\\!\\?])', '', text)
    text = re.sub('[\\.\\:\\!\\?]$', '', text)
    return text

def find_best_match(text: str, candidate_texts: list[str]) -> tuple[int, str]:
    """
    Given "text" and a list of possible matches ("candidate_texts"), return the
    index of the best match and the match itself. We assume there is a match.

    :param text: The text to match.
    :param candidate_texts: The list of candidate texts to match against.
    :return: A tuple of the index of the best match and the match itself.
    """
    candidates_normalised = [normalise_text_for_comparison(candidate) for candidate in candidate_texts]
    text_normalised = normalise_text_for_comparison(text)
    for i, candidate in enumerate(candidates_normalised):
        if candidate == text_normalised:
            return (i, candidate_texts[i])
    logger.info('No good match! Using substring matching...')
    best_match = ''
    match_index = -1
    best_substring_length = 0
    for i, candidate in enumerate(candidates_normalised):
        substring_match = SequenceMatcher(None, text_normalised, candidate).find_longest_match(0, len(text_normalised), 0, len(candidate))
        substring = text[substring_match.a:substring_match.a + substring_match.size]
        if len(substring) > best_substring_length:
            best_match = candidate_texts[i]
            match_index = i
            best_substring_length = len(substring)
    assert match_index != -1, f'Failed to find a match for {text} in {candidate_texts}.'
    return (match_index, best_match)


ARGS = {
    "text": "Hello World!",
    "candidate_texts": [
        "hello world",
        "goodbye world",
        "hello there"
    ]
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_prompts_dir Ran
ran; our contract draft was wrong, not the code
microsoft/RadFact/src/radfact/paths.py
code served (permissive licence) · get_code("d1c6dce3ee15fa92")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned PosixPath. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

from pathlib import Path

ARGS = {
    "task": "fact_checking"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type PosixPath — a PosixPath output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a PosixPath output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
normalise_text_for_comparison Ran
ran; our contract draft was wrong, not the code
microsoft/RadFact/src/radfact/llm_utils/text_utils.py
code served (permissive licence) · get_code("0cefd97cbed67d49")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import re

def normalise_text_for_comparison(text: str) -> str:
    """
    Normalise a string for comparison by removing whitespace, dashes, and punctuation.

    :param text: The text to normalise.
    :return: The normalised text.
    """
    text = text.lower().strip()
    text = re.sub('\\s+', ' ', text)
    text = re.sub('[\\-–—]', '', text)
    text = re.sub('\\s*(?<=[\\.\\:\\!\\?])', '', text)
    text = re.sub('[\\.\\:\\!\\?]$', '', text)
    return text

ARGS = {
    "text": "Hello  World---Test! Multiple   spaces:and-dashes?"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
token_to_json Ran
ran; our contract draft was wrong, not the code
microsoft/RadFact/src/radfact/azure_utils/auth.py
code served (permissive licence) · get_code("8e0e246be6c70e29")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import base64
import json
from typing import Any

def token_to_json(token: str) -> Any:
    """Converts an Azure access token to its underlying JSON structure.

    :param token: The access token.
    :return: The JSON object that is stored in the token.
    """
    base64_meta_data = token.split('.')[1].encode('utf-8') + b'=='
    json_bytes = base64.decodebytes(base64_meta_data)
    json_string = json_bytes.decode('utf-8')
    return json.loads(json_string)


# Create a valid Azure-like access token
# Azure tokens are JWT-like with format: header.payload.signature
# We need to create a valid base64-encoded payload in the middle part

# Create a JSON payload to encode
payload = {
    "aud": "https://management.azure.com/",
    "iss": "https://sts.windows.net/12345678-1234-1234-1234-123456789012/",
    "iat": 1234567890,
    "exp": 1234571490
}

# Convert to JSON string and then to base64
payload_json = json.dumps(payload)
payload_bytes = payload_json.encode('utf-8')
payload_base64 = base64.b64encode(payload_bytes).decode('utf-8').rstrip('=')

# Create header and signature (simple placeholders)
header_json = json.dumps({"alg": "RS256", "typ": "JWT"})
header_bytes = header_json.encode('utf-8')
header_base64 = base64.b64encode(header_bytes).decode('utf-8').rstrip('=')

signature = "valid_signature_placeholder"

# Construct the full token in JWT format
token = f"{header_base64}.{payload_base64}.{signature}"

ARGS = {
    "token": token
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 4 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Radiology reporting is a complex task requiring detailed medical image understanding and precise language generation, for which generative multimodal models offer a promising solution. However, to impact clinical practice, models must achieve a high level of both verifiable performance and utility. We augment the utility of automated report generation by incorporating localisation of individual findings on the image - a task we call grounded report generation - and enhance performance by incorporating realistic reporting context as inputs. We design a novel evaluation framework (RadFact) leveraging the logical inference capabilities of large language models (LLMs) to quantify report correctness and completeness at the level of individual sentences, while supporting the new task of grounded reporting. We develop MAIRA-2, a large radiology-specific multimodal model designed to generate chest X-ray reports with and without grounding. MAIRA-2 achieves state of the art on existing report generation benchmarks and establishes the novel task of grounded report generation.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2406.04449")
get_code_for_paper("2406.04449")
have("2406.04449")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2406.04449.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2406.04449.svg)](https://syntology.ai/paper/2406.04449)

Connect an agent — have() is free.