BENCHMARKS FLOWING
Deux benchmarks : qualité de rédaction et détection des erreurs
Sur 7 disciplines, 130 articles distincts et 280 positions d’écriture fixes, l’autocomplétion avec Evidence obtient un avantage transversal net de 31 victoires ; Evidence Card détecte 91,4 % des six types d’erreurs explicites.
EVIDENCE CARD · TEST CONTRÔLÉ · REVUE INDÉPENDANTE
Nous avons introduit des erreurs nettes. Evidence Card en a détecté 91,4 %.
Rappel des erreurs Red
32/35 confirmés ; 33/35 en comptant le cas à revoir.
Précision des cartes Red
34/35 confirmées · 1 à revoir · 0 rejetée
Exactitude des contrôles Green
11/11 contrôles Green correctement traités ; 0/11 fausses alertes Red.
PRÉCISION YELLOW · CAS NATURELS
6/8 confirmées par une revue indépendante
Les tâches et les dénominateurs diffèrent ; les résultats sont présentés séparément.
MESURES PUBLIÉES
Rappel, précision des cartes, contrôles et taux d’achèvement
Chaque mesure conserve son propre dénominateur. Les échecs techniques sont indiqués séparément et ne comptent pas comme « aucun problème ».
Rappel des erreurs Red
91,4 %32/35 confirmés ; 33/35 en comptant le cas à revoir.
Précision des cartes Red
97,1 %34/35 confirmées · 1 à revoir · 0 rejetée
Exactitude des contrôles Green
100,0 %11/11 contrôles Green correctement traités ; 0/11 fausses alertes Red.
Tests terminés
95,8 %46/48 positions terminées ; 2 échecs techniques indiqués à part, jamais comptés comme « aucun problème ».
RÉSULTATS PAR TYPE D’ERREUR
Résultats des six types d’erreur
Chiffres, direction, causalité, objet, portée et théorie ou définition sont présentés séparément.
Foncé = confirmé ; clair = à revoir
Chiffres
5/6Aucun échec technique
Direction ou polarité
6/6Aucun échec technique
Causalité
4/51 échec technique indiqué à part
Mauvais objet
5/6Aucun échec technique
Champ d’application
6/6Aucun échec technique
Théorie ou définition
6/6Aucun échec technique
PRÉCISION YELLOW · CAS NATURELS
6/8 confirmées par une revue indépendante
Cartes Yellow avant et après ajustement
La précision stricte est passée de 31,3 % à 75,0 %.
Avant l’ajustement, 16 cartes ont été émises : 5 confirmées, 1 à revoir et 10 rejetées. La relance complète en a produit 8 : 6 confirmées, 0 à revoir et 2 rejetées. 50/51 positions ont été terminées ; 1 échec technique a été indiqué à part.
Relance complète et revue indépendante de chaque carte terminées48 CAS PUBLICS
Entrées, cartes et résultats de revue pour chaque cas
Chaque cas comprend l’affirmation testée, la source figée, la carte produite et la revue indépendante. Les omissions et les échecs techniques restent dans le relevé.
REDChiffresDétection confirmée溶液法原位大面积制备钙钛矿光电薄膜成膜的同步辐射可视化结晶过程研究
该研究报道,单个器件面积约15 cm²的大面积全柔性三维钙钛矿太阳能电池转换效率提升至15.23%。
Ouvrir le cas ↓
Affirmation testée
该研究报道,单个器件面积约15 cm²的大面积全柔性三维钙钛矿太阳能电池转换效率提升至15.23%。
Source figée
转换效率提升至5.23%(单个器件面积约15 cm2)
Carte produite · RED
目标将该研究中约15 cm²全柔性三维钙钛矿太阳能电池的转换效率写为15.23%,而原文对同一器件报告的是5.23%,构成直接数值矛盾。
“基于各层均采用卷对卷 全溶液狭缝涂布方法制备的大面积全柔性三维钙钛矿薄膜太阳能电池转换效率提升至5.23% (单个器件面积 约15 cm2)”— 溶液法原位大面积制备钙钛矿光电薄膜成膜的同步辐射可视化结晶过程研究
Revue indépendante · valid
TARGET QUOTE声称'单个器件面积约15 cm²的大面积全柔性三维钙钛矿太阳能电池转换效率提升至15.23%',SOURCE QUOTE对同一器件(单个器件面积约15 cm²、大面积全柔性三维钙钛矿薄膜太阳能电池)报告转换效率为'5.23%'。两个数值15.23%与5.23%在相同实体(约15 cm²大面积全柔性三维钙钛矿太阳能电池)、相同关系(转换效率)上存在直接数值矛盾,构成直接反驳。
REDChiffresDétection confirméeGenome-Scale CRISPR-Cas9 Knockout Screening in Human Cells
The GeCKO library targeted 18,080 human genes with 46,751 unique guide sequences.
Ouvrir le cas ↓
Affirmation testée
The GeCKO library targeted 18,080 human genes with 46,751 unique guide sequences.
Source figée
targeting 18,080 genes with 64,751 unique guide sequences
Carte produite · RED
The stated 46,751 unique guide sequences contradicts the source's 64,751 for the same GeCKO library. The source agrees with the stated 18,080 targeted genes.
“We show that lentiviral delivery of a genome-scale CRISPR-Cas9 knockout (GeCKO) library targeting 18,080 genes with 64,751 unique guide sequences enables both negative and positive selection screening in human cells.”— Genome-Scale CRISPR-Cas9 Knockout Screening in Human Cells
Revue indépendante · valid
The TARGET QUOTE states '46,751 unique guide sequences' for the GeCKO library, while the SOURCE QUOTE states '64,751 unique guide sequences' for the same GeCKO library targeting the same 18,080 genes. This is a direct numerical contradiction on the exact same entity (GeCKO library), same relation (number of unique guide sequences), and same scope. The source explicitly confirms the 18,080 genes figure but contradicts the 46,751 guide sequences figure. The proposed 'red' judgment is correct.
REDChiffresDétection confirméeEmpagliflozin, Cardiovascular Outcomes, and Mortality in Type 2 Diabetes
EMPA-REG OUTCOME treated a total of 8,020 patients and followed them for a median of 3.1 years.
Ouvrir le cas ↓
Affirmation testée
EMPA-REG OUTCOME treated a total of 8,020 patients and followed them for a median of 3.1 years.
Source figée
A total of 7020 patients were treated (median observation time, 3.1 years)
Carte produite · RED
The EMPA-REG OUTCOME report states that 7,020 patients were treated, not 8,020.
“A total of 7020 patients were treated (median observation time, 3.1 years).”— Empagliflozin, Cardiovascular Outcomes, and Mortality in Type 2 Diabetes
Revue indépendante · valid
The target quote states '8,020 patients' while the source quote explicitly states '7020 patients were treated.' This is a direct numerical contradiction on the identical claim of patient count in the EMPA-REG OUTCOME trial. The entity (EMPA-REG OUTCOME trial), relation (patient count), and scope (total treated patients) match exactly, with only the number differing. The proposed color red is correct because the source directly contradicts the target proposition on a material point.
REDChiffresDétection confirméeTabPFN: A Transformer That Solves Small Tabular Classification Problems in a Second
The paper's small-task setting allows up to 10,000 training examples, 100 numerical features, and 10 classes.
Ouvrir le cas ↓
Affirmation testée
The paper's small-task setting allows up to 10,000 training examples, 100 numerical features, and 10 classes.
Source figée
≤ 1 000 training examples, ≤ 100 purely numerical features and ≤ 10 classes
Carte produite · RED
The paper defines its small-task setting as at most 1,000 training examples, not 10,000.
“has thus learned to solve novel small tabular classification tasks (≤ 1 000 training examples,≤ 100 purely numerical features without missing values and≤ 10 classes)”— TabPFN: A Transformer That Solves Small Tabular Classification Problems in a Second
Revue indépendante · valid
The TARGET QUOTE states 'up to 10,000 training examples,' while the SOURCE QUOTE explicitly states '≤ 1 000 training examples.' This is a direct numerical contradiction on the same entity (training examples), same relation (upper bound on small-task setting), same scope (the paper's definition of small-task setting). The source directly contradicts the target proposition that the limit is 10,000 by specifying it is 1,000.
REDChiffresDétection confirméeThe Progress, Challenges, and Perspectives of Directed Greybox Fuzzing
The survey investigates 24 state-of-the-art fuzzers closely related to directed greybox fuzzing.
Ouvrir le cas ↓
Affirmation testée
The survey investigates 24 state-of-the-art fuzzers closely related to directed greybox fuzzing.
Source figée
investigation of 42 state-of-the-art fuzzers that are closely related to DGF
Carte produite · RED
The survey explicitly reports investigating 42 state-of-the-art fuzzers closely related to DGF, not 24.
“Based on the investigation of 42 state-of-the-art fuzzers that are closely related to DGF, we conduct the first in-depth study to summarize the empirical evidence on the research progress of DGF.”— The Progress, Challenges, and Perspectives of Directed Greybox Fuzzing
Revue indépendante · valid
The target quote claims '24 state-of-the-art fuzzers closely related to directed greybox fuzzing.' The source quote explicitly states '42 state-of-the-art fuzzers that are closely related to DGF' (where DGF = directed greybox fuzzing). The entity (fuzzers closely related to directed greybox fuzzing), relation (investigation of), and scope (state-of-the-art) match exactly; the only difference is the number (24 vs. 42), which is a direct numerical contradiction on a material point. The proposed color 'red' is correct.
REDChiffresNon détectéEvaluating the Impact of Brazil's Bolsa Família: Cash Transfer Programs in Comparative Perspective
The review describes Bolsa Família as benefiting roughly 17 million families.
Ouvrir le cas ↓
Affirmation testée
The review describes Bolsa Família as benefiting roughly 17 million families.
Source figée
benefiting roughly 11 million families
Aucune carte produite
REDDirection ou polaritéDétection confirméeEmpagliflozin, Cardiovascular Outcomes, and Mortality in Type 2 Diabetes
In EMPA-REG OUTCOME, cardiovascular death was significantly higher with empagliflozin than with placebo.
Ouvrir le cas ↓
Affirmation testée
In EMPA-REG OUTCOME, cardiovascular death was significantly higher with empagliflozin than with placebo.
Source figée
significantly lower rates of death from cardiovascular causes (3.7%, vs. 5.9% in the placebo group)
Carte produite · RED
The claim that cardiovascular death was higher with empagliflozin contradicts the account of the same trial’s cardiovascular results, which explicitly reports a lower rate of death from cardiovascular causes.
“Cardiovascular Outcomes The primary outcome that was reported previ ously showed a significantly lower risk of 3-point MACE in the pooled empagliflozin group than in the placebo group, a difference that was driven by a lower rate of death from cardiovas cular causes.23”— Empagliflozin and Progression of Kidney Disease in Type 2 Diabetes
Revue indépendante · valid
The target quote explicitly claims cardiovascular death was 'significantly higher with empagliflozin than with placebo' in EMPA-REG OUTCOME. The source quote directly contradicts this by stating there was 'a lower rate of death from cardiovascular causes' in the empagliflozin group, which 'drove' the composite benefit. Both quotes concern the same trial (EMPA-REG OUTCOME), the same outcome (cardiovascular death/mortality), the same comparison (empagliflozin vs. placebo), and the same directionality claim—but with opposite directions. The contradiction is material and unambiguous: 'higher' vs. 'lower' for the identical endpoint in the identical trial.
REDDirection ou polaritéDétection confirméeEvaluating the Impact of Brazil's Bolsa Família: Cash Transfer Programs in Comparative Perspective
The review concludes that Bolsa Família increased inequality and extreme poverty.
Ouvrir le cas ↓
Affirmation testée
The review concludes that Bolsa Família increased inequality and extreme poverty.
Source figée
helped reduce inequality and extreme poverty
Carte produite · RED
The review attributes 21 percent of Brazil’s 1995–2004 Gini decline to Bolsa Família, contradicting the claimed increase in inequality.
“They found that the Gini index for Brazil fell by 4.7 percent from 1995 to 2004 and that Bolsa Familia was responsible for 21 percent of that fall (before 1995 there were no CCT programs in Brazil, so there is confidence that 1995 is a baseline year).”— Evaluating the Impact of Brazil's Bolsa Família: Cash Transfer Programs in Comparative Perspective
Revue indépendante · valid
The TARGET QUOTE states 'The review concludes that Bolsa Família increased inequality.' The SOURCE QUOTE from the same review states that Bolsa Família was responsible for '21 percent of that fall' in the Gini index, where 'fall' means a decrease in inequality. A decrease in the Gini index signifies reduced inequality. Therefore, the source directly contradicts the target proposition: the review concludes Bolsa Família decreased inequality (contributed to 21% of the Gini decline), not increased it. The scope matches—both concern the review's conclusion about Bolsa Família's effect on inequality.
Portée et limites
- 1Ce test contrôlé mesure la capacité à repérer des erreurs explicites. Il n’estime pas leur fréquence dans de vrais articles.
- 2Certains articles sources ont aussi servi lors de premiers essais. Ces résultats conviennent aux tests de régression et au rappel contrôlé, pas à une estimation de prévalence sur échantillon aléatoire.
- 3La version ajustée n’a émis que 8 cartes Yellow. Le dénominateur 6/8 doit rester visible ; 75 % n’est pas une estimation de la précision globale.