Do Factuality Methods Make LLMs More Factual? Reproducibility and Fair Comparison

Pratuat Amatya; Vinay Setty; V Venktesh; Avishek Anand

cs.CLOct 13, 2026

Do Factuality Methods Make LLMs More Factual? Reproducibility and Fair Comparison

Authors: Pratuat Amatya, Vinay Setty, V Venktesh, Avishek Anand

Organizations: Factiverse AS · University of Stavanger · Stockholm University · TU Delft

Abstract

Large Language Models (LLMs) are aligned to human preferences, after pre-training and supervised fine-tuning steps. However, standard RL alignment steps such as RLHF do not focus on factuality of LLMs, hence they may generate non-factual information, which could result in unintended consequences. In response, a number of factuality alignment approaches have been proposed in the literature which claim to improve factuality of LLMs on their own benchmarks. However, these methods have not been reproduced independently and are also not comparable since each uses a different base model, training data, evaluation set, judge, and metric. In this paper, we present the first independent reproduction of five factuality alignment methods, which span on-policy RL (PPO, GRPO, and Dr.~GRPO) and offline preference optimization (DPO and KTO), and compare them on a common benchmark with common evaluation metrics. We found that none of these methods reproduces in full, and on the common benchmark none of the methods significantly outperforms the base model. On closer inspection, the rewards optimize for format, length, or brevity rather than factuality and relevance, whereas the choice between on-policy and offline training matters little. Furthermore, the evaluation metrics and benchmarks are also part of the problem since they do not measure relevance of text to the prompt. Since there is no ground truth associated with the prompts, the reliance on LLM judges and retrieval for verification also makes the evaluation fragile. We release our code, data, and checkpoints.

Figures & tables

Panel of Figure 1 — Figure 1: Published gains vs. reproduced. (a) Gain over each method’s own baseline and as we measure it in the original setup, with released checkpoints as circles. (b) Change over the base model on the common benchmark with 95% bootstrap intervals.

Panel of Figure 2 — Figure 2: The five alignment methods and our reproductions. Gray boxes show shared stages, and colored boxes show differences between methods. Flame and FactAlign use offline preference optimization; the other three methods use on-policy RL with explicit rewards. The final column records the main deviation in each reproduction. Further details appear in Section 4 and Appendix A .

Quantity	Paper	Ours
Flame
Bio FActScore, $\textsc{Sft}^{\text{f}}{+}\textsc{Dpo}^{\text{f}}$	47.9	54.2	✓
AlpacaEval win rate vs. SFT ${+}$ DPO	51.2	37.7	✗
FactAlign , $\text{F1}@100$
released checkpoint vs. base, paired	—	$+2.68$	✓
our three-iteration run vs. its baseline	$+24.8$	$-2.64$	✗

Table 1: Published and reproduced results under each method’s protocol, in points unless noted. Compare within rows only. ✓: reproduced; ✗: not reproduced;

\approx

: same direction, different magnitude.

SimpleQA ( $n=1{,}000$ )
Method	Acc.	Abst.	Acc $\mid$ att.	F
base (untrained)	.034	.812	.181	.057
Flame	.032	.851	.215	.056
FactAlign	.032	.813	.171	.054
KnowRL	.032	.822	.180	.054
FactGRPO	.052	.414	.089	.066

Table 2: Short-form evaluation, one judge ( gpt-4o ) across every method. Abst. is the share of questions the model declined; Acc

\mid

att. is accuracy over attempted questions only, which is the quantity a method can raise by declining more.

Biography ( $n=490$ , $K=20$ )
		VeriScore				VeriScore + Relevance
Method	#Claims	#Sup.	P	R	F1	#Sup.	P	R	F1
base (untrained)	27.6	12.0	0.351	0.472	0.370	10.6	0.283	0.434	0.325
Flame	23.5	10.6	0.338	0.436 †	0.356 †	9.7	0.294	0.407 †	0.324
FactAlign	27.1	11.7	0.343	0.468	0.368	10.4	0.284	0.431	0.326
KnowRL	29.8	11.3	0.281 †	0.440 †	0.325 †	10.0	0.237 †	0.409 †	0.291 †

Table 3: Pooled results for long-form factuality under VeriScore and VeriScore with relevance filtering evaluation. #Claims is the mean number of extracted claims per response and #Sup. the mean number of supported claims; bold marks the highest value per column within each dataset. † marks a significant difference from base (paired bootstrap test, Holm-corrected

p<0.05

Panel of Figure 3 — Figure 3: Mean FactGRPO reward as a function of the number of filler (supported but irrelevant) claims in a long-form biography response, for the rerun with the full reward and for the first run without the relevance term. Bins beyond 14 hold one or two responses and are not shown.

Appendix figures & tables26 assets

Supplementary material from the paper’s appendix.

Appendix

Llama-2 70B	Bio	A-Fact	FAVA	Win%
SFT	53.9	44.3	54.7	26.1
SFT + DPO	51.6	43.7	51.5	50.0
Sft ${}^{\text{f}}$	54.6	45.5	54.2	19.3
Flame	54.2	46.2	50.9	37.7

Appendix

Table 4: Flame at 70B: FActScore on 183 biographies, 253 Alpaca-Fact prompts, and FAVA; AlpacaEval win rate against SFT+DPO using gpt-4-turbo . Published biography scores are 44.7, 42.3, 49.5, and 47.9 for the four rows.

Method	Code	Training data	Checkpoints	Reward or judge	Released code vs. paper
Flame	✗	prompts only; SFT and DPO data are self-generated	✗	reward models ✗	no code to compare
FactAlign	✓	self-generated; general-domain mixture?	✓	SAFE pipeline ✓	evaluates a LongFact variant regenerated in 2026, which offsets every absolute number by about 19 points
KnowRL	✓	✓	✓	API judge, snapshot drifts	cold-start stage via LLaMA-Factory with the released data; the released RL configuration sets learning rate $10^{-4}$ and rank 128 where the paper gives $10^{-5}$ and 256, with the base model left as a placeholder; the fact reward receives the response without its opening reasoning tag
FactGRPO	✓	SFT data ✓; RL prompts recovered from the DPO pairs (3,706 of 4k)	?	programmatic ( VeriScore ) ✓	$\lambda_{\text{len}}=0$ in the released configuration
Faith	✓	✓, with the estimator’s predicted state in every prompt	?	reward model trained from released code	one epoch for the reward model instead of two and 0.45 PPO epochs instead of about two in the Llama-3 configurations; the rectifier ships a Mistral configuration but no training data (provided by the authors on request) and no Llama-3 configuration

Appendix

Table 5: What each method releases and where the released code departs from its paper. ✓ released, ✗ not released, partial when only a part is usable, and ? when it is unclear.

	Llama-3-8B-Inst.	FactAlign
claims	46.00	45.22
supported	30.54	32.39
not supported	14.71	11.98
precision	0.675	0.730
$\text{F1}@100$	0.4145	0.4413

Appendix

Table 6: FactAlign ’s released checkpoint against its base, both rated in one paired SAFE pass over 190 prompts. Counts are per response.

Run	1	2	3	4	5
KnowRL , end to end	0.2	0.2	0.1	0.2	0.2
KnowRL-Zero , paper hyperparameters	7.2	14.2	33.7	93.1	96.8
+ repaired fact input	7.4	15.2	82.2	98.3	98.3
released hyperparameters, raw model	65.2	99.6	99.9	100.0	100.0
released hyperparameters, after cold start	0.9	15.7	85.2	99.2	98.4

Appendix

Table 7: KnowRL refusal rate in percent by fifth of training. The method as specified does not abstain, whereas every run at the released learning rate or without the paper’s cold start collapses, and repairing the input of the fact reward does not prevent the collapse, which locates the cause in the cold-start stage and the learning rate rather than in the fact signal.

	Base distill	KnowRL
Reported accuracy	1.92	4.56
our measurement	2.00	5.71
Rejected	10.2%	93.0%
Correct answers, of 500	9	2

Appendix

Table 8: The released KnowRL checkpoint on 500 SimpleQA items under the authors’ protocol. The reported gain reproduces, but rejection rises from 10% to 93% and the number of correctly answered questions falls.

SimpleQA ( $n=1{,}000$ )
Method	Acc.	Abst.	Acc $\mid$ att.	F
without the relevance term	.053	.401	.088	.066
full reward	.052	.414	.089	.066
TruthfulQA ( $n=817$ )
without the relevance term	.237	.355	.368	.289
full reward	.231	.368	.366	.284

Appendix

Table 9: Short-form results of the two FactGRPO runs on the common benchmark, graded by gpt-4o . The first run was trained without the relevance term of the reward, and the rerun with the full reward is the variant used in the main tables.

	Paper	Ours
Base model	Llama-3.1-8B-Inst.	same
SFT data	3k synthetic	their released 3k
RL prompts	4k synthetic	3,706 recovered
Rollouts	4	4
Learning rate	1e-6	1e-6
Objective	Dr. GRPO	Dr. GRPO

Appendix

Table 10: Our FactGRPO run against the published configuration of Chen et al. (2026) . The last three rows are the deviations that matter.

	Biography		Alpaca-Fact
	$P$	$S$	$P$	$S$
Base	35.8	12.1	61.9	20.0
${+}$ our GRPO	36.9	10.9	64.7	18.1
Paper, base	25.9	10.3	60.2	30.7
Paper, ${+}$ GRPO	47.1	14.8	79.7	36.9

Appendix

Table 11: VeriScore precision (

P

) and detail level (

S

) on the long-form benchmarks, base model against our 400-step run, with the published effect for reference.

Biography ( $n=488$ , $K=20$ )
		VS				VS+Rel
Method	#Claims	#Sup.	P	R	F1	#Sup.	P	R	F1
without the relevance term	25.6	10.5	0.360	0.480	0.399	9.4	0.316	0.438	0.357
full reward	25.4	10.1	0.342	0.466	0.384	9.1	0.301	0.424	0.343
Alpaca-Fact ( $n=246$ , $K=19$ )
without the relevance term	26.5	18.2	0.631	0.699	0.638	13.4	0.471	0.573	0.491

Appendix

Table 12: FactGRPO with and without the relevance reward, evaluated on shared prompts before and after relevance filtering.

K

matches Table 3 .

Dataset	Est.	Read	Est.	Read	Agree
	SFT		SFT $+$ PPO
NQ-Open	.297	.303	.295	.298	83.2%
SciQ	.585	.584	.598	.591	84.1%
TriviaQA	.662	.662	.659	.658	90.1%
WebQuestions	.342	.341	.347	.346	82.6%
overall					87.5%

Appendix

Table 13: Faith with predicted (Est.) or released, precomputed (Read) knowledge states. Agree measures state agreement. Both models use Meta-Llama-3-8B (Section 4 ); Est. follows Section 3.4 of the original paper.

Dataset	Sft	Faith	$\Delta$	$\Delta$
	Ours, precision			Full system
TriviaQA	74.15	73.56	$-0.60$	$+13.39$
SciQ	65.74	67.13	$+1.39$	$+8.43$
NQ-Open	42.51	42.57	$+0.06$	$+16.72$
WebQuestions	43.27	43.74	$+0.47$	$+1.53$
Mean			$+0.33$	$+10.02$

Appendix

Table 14: Faith on its own four validation sets, scored in one harness. Precision is over the questions the estimator predicts as known. The last column is the gain of the paper’s full system on Llama-3-8B, whereas our delta compares our own SFT stage with our own PPO model, which corresponds to the paper’s ablation without the retrieval stage, reported in Table 1 .

SimpleQA ( $n=1{,}000$ )
Method	Acc.	Abst.	Acc $\mid$ att.	F
base (untrained)	.034	.812	.181	.057
Flame	.029	.860	.207	.051
KnowRL	.029	.831	.172	.050
FactGRPO	.054	.416	.092	.068
Faith	.046	.001	.046	.046

Appendix

Table 15: SFT stage results for short-form evaluation.

Biography ( $n=491$ , $K=20$ )
Method	Sent.	$S$	$U$	Prec.	$\text{F1}@K$
base (untrained)	13.6	12.02	15.60	.351	.370
Flame	11.6	10.30	12.35	.334	.349
KnowRL	16.0	11.33	19.31	.270	.316
FactGRPO	12.7	10.33	15.15	.353	.394
Faith	1.0	0.39	0.21	.373	.037

Appendix

Table 16: Long-form results of the supervised stages alone, scored in the same VeriScore configuration and with the same fixed

K

as Table 3 .

SimpleQA ( $n=1{,}000$ )
Method	Acc.	Abst.	Acc $\mid$ att.	F
Faith SFT	.055	.001	.055	.055
Faith SFT + state	.046	.001	.046	.046
Faith full	.058	.001	.058	.058
Faith full + state	.049	.000	.049	.049
TruthfulQA ( $n=817$ )

Appendix

Table 17: Short-form results of Faith with and without the estimator-predicted knowledge state in the prompt, for the supervised stage and the full pipeline.

Biography ( $n=495$ , $K=20$ )
Method	Sent.	$S$	$U$	Prec.	$\text{F1}@K$
Faith SFT	1.0	0.54	0.24	.446	.050
Faith SFT + state	1.0	0.39	0.21	.374	.037
Faith full	1.0	0.52	0.22	.444	.048
Faith full + state	1.0	0.39	0.21	.377	.037
Alpaca-Fact ( $n=253$ , $K=19$ )

Appendix

Table 18: Long-form results of the same four Faith variants, with the fixed

K

of Table 3 .

	Reward
Generations	Flame	FactAlign	KnowRL	FactGRPO	Faith
base (untrained)	$1.13$	$-1.15$	$-0.392$	$-1.000$	$3.85$
Flame	$-0.09$ †	$-0.25$ †	$+0.050$ †	$0.000$	$+2.16$ †
FactAlign	$-0.01$	$+0.05$ †	$+0.026$	$0.000$	$+0.03$
KnowRL	$-0.12$ †	$+0.03$	$+0.004$	$0.000$	$-0.32$
FactGRPO	$-0.40$ †	$-2.24$ †	$-0.349$ †	$+1.028$ †	$-3.97$ †

Appendix

Table 19: Mean reward of each method’s generations under each method’s reward, as the difference from the base model’s generations, whose absolute value is in the first row. Rewards are per thousand words ( Flame , FactAlign , Faith ) or per response ( KnowRL on short-form and FactGRPO on long-form prompts). Bold marks a method under its own reward, and † a significant difference from the base row (paired bootstrap, Holm-corrected

p<0.05

). Faith generations are left out, since one-word answers make per-word values not comparable.

SimpleQA ( $n=1{,}000$ )
	gpt-4o		gpt-oss-120b
Method	Acc.	F	Acc.	F
base (untrained)	.034	.057	.030	.051
Flame	.032	.056	.030	.052
FactAlign	.032	.054	.029	.049
KnowRL	.032	.054	.031	.052

Appendix

Table 20: The short-form evaluation under both judges. Acc. is accuracy over all questions, F the harmonic mean of accuracy and accuracy-given-attempted. Absolute values shift systematically between judges; only differences within a column are meaningful. Items that the open judge left ungraded, up to 58 of 817 on TruthfulQA, are excluded from its rates.

Question	Response ( <answer> span)	$R_{\mathrm{fact}}$
Tell me a bio of Diego Suarez (garden designer).	I’m not sure if I have any information on a garden designer by the name of Diego Suarez. However, I can provide some general information about garden designers … Garden designers are professionals who design and plan gardens, taking into account factors such as the climate, soil type, and desired aesthetic. … Some well-known garden designers include Gertrude Jekyll, Russell Page, and Piet Oudolf. … I apologize, but I don’t have any information on a garden designer by the name of Diego Suarez.	0.93
Tell me a bio of Shaolin (humorist).	I’m not sure if I have any information about a well-known humorist by the name of Shaolin. However, I can provide some general information about the Shaolin Temple in China … The Shaolin Temple is located in Dengfeng, Henan Province, China … has a rich history that dates back to the 5th century, and it’s been a major center of Buddhist learning … particularly Shaolin Kung Fu …	0.84
Tell me a bio of Alexandre Hohagen.	Alexandre Hohagen is a name that combines a French first name with a German surname. The name Alexandre is derived from the Greek name Alexander , which means “defender of the people.” The surname Hohagen is of German origin … Unfortunately, I don’t have any specific information about a person named Alexandre Hohagen. …	0.75
Tell me a bio of Kubota Beisen.	I’m not familiar with a well-known artist by the name of Kubota Beisen. … However, I can provide some general information about Japanese art and artists from the same time period . The Edo period (1603–1868) and the Meiji period (1868–1912) were significant periods in Japanese art history … Some notable Japanese artists from these periods include Hokusai … Hiroshige … and Yokoyama Taikan …	0.74
Tell me a bio of Rangi Te Kanawa.	Rangi Te Kanawa is a New Zealand singer, but I couldn’t find much information about their background … you might be thinking of Kiri Te Kanawa . Kiri Te Kanawa is a prominent New Zealand soprano … She was born in 1944 in Gisborne, New Zealand … and has since performed with many major opera companies … including the Metropolitan Opera in New York and the Royal Opera in London. …	0.67

Appendix

Table 21: FactGRPO responses from the first run that drift away from the requested subject yet earn a high factuality reward ( <answer> span only; … marks omitted text). Italics mark the substituted subject: a profession, a homonym, the name’s etymology, an art-historical period, and a similarly named person.

R_{\mathrm{rel}}

is 0.1 only when an LLM judge prefers the response to the reference response, and 0 otherwise;

R_{\mathrm{fact}}=S/(T{+}1)

over

T

extracted claims,

S

of them supported. Because

R_{\mathrm{rel}}\leq 0.1

, verifiable but off-topic content dominates the total reward.

Panel of Figure 4 — Figure 4: Distribution of the number of filler (supported but irrelevant) claims within long-form responses for the Biography dataset. The y-axis is logarithmic, and a missing point means a count of zero.

Panel of Figure 5 — Figure 5: Distribution of the number of hallucinated (relevant but unsupported) claims within long-form responses for the Biography dataset. The y-axis is logarithmic, and a missing point means a count of zero.

Biography ( $n=492$ )
	$K$ (biography / Alpaca-Fact)
Method	5 / 7	14 / 15	20 / 18	23 / 21	29 / 24
base	.442	.398	.371	.357	.331
Flame	.435	.394	.367	.353	.326
FactAlign	.427	.384	.357	.342	.315
KnowRL	.421	.378	.350	.336	.311

Appendix

Table 22:

\text{F1}@K

of the same responses under the per-domain

K

produced by different model pools scored in this study, given as biography / Alpaca-Fact. From left to right, the pools are a twelve-model pool with five Faith variants, a ten-model pool, a nine-model pool, the six-run pool of the earlier scoring pass, and a single model scored alone. The responses are those of the earlier scoring pass, with the first FactGRPO run. Precision does not depend on

K

Question	Response claims	Comment
Tell me a bio of Mindy Smith.	[ FactGRPO ] ✗ Mindy Smith has released several singles throughout her career. Below is a list of some of her notable releases: • - “One Moment More” • - “It’s That Time Again” • - “Come to Jesus” • Mindy Smith has been performing at festivals and venues across the United States. …	VeriScore extracts bare song titles as check-worthy claims and labels them supported.
Tell me a bio of Jessie Mae Brown Beavers.	[ FactAlign ] ✗ Jessie Mae Brown Beavers was active in the Civil Rights Movement in Mississippi. • Jessie Mae Brown Beavers was a community leader. • Jessie Mae Brown Beavers was an community leader. • Jessie Mae Brown Beavers was an activist. …	Near-duplicate claims are each counted as separately supported, inflating the factuality score.
Tell me a bio of Diego Suarez (garden designer).	[ FactGRPO ] • Garden designers design and plan gardens. • Garden designers consider the climate when designing gardens. • Garden designers consider the soil type when designing gardens. • Garden designers aim to achieve a desired aesthetic in their designs. …	Generic filler claims are factually supported but irrelevant to the question. FactGRPO exploits such claims to hack the reward (Table 21 ).

Appendix

Table 23: Examples of limitations in claim-level factuality evaluation. Non-claims, repeated claims, and irrelevant claims are each counted as supported, inflating the factuality score. A cross (✗) in place of the bullet marks claims that VeriScore labels not supported (all other claims are labeled supported), and typeface marks the relevance judgment ( bold : relevant, italic : irrelevant).

Method	#Claims	#Sup.	P	#Sup.	P	Agree.
		Web search		Wikipedia
Biography ( $n=489$ )
base (untrained)	27.6	12.0	.351	5.5	.140	.729
Flame	23.6	10.6	.338	5.1	.149	.729
FactAlign	27.1	11.7	.344	5.5	.141	.734
KnowRL	29.8	11.3	.281 †	5.1	.116 †	.760

Appendix

Table 24: Comparison between evidence sources for verification. Biography claims of each method verified against web search vs against the subject’s Wikipedia article with the same verifier, on the 489 prompts with an article that every method answered. #Sup. is the mean number of supported claims per response, P the mean per-response precision, and Agreement the share of claims with the same verdict under both sources. † marks a significant difference from the base row (paired bootstrap, Holm-corrected

p<0.05

Web support vs. VeriScore
	$n$	Agr.	$\kappa$
first annotator	261	.797	.58
second annotator	289	.751	.49
between annotators	261	.778	.50
consensus	203	.842	.65
Relevance vs. LLM labels

Appendix

Table 25: Agreement of the human audit with the automatic labels. Agr. is the share of claims with the same label and

\kappa

Cohen’s kappa. Support rows count Cannot tell from evidence as unsupported, as the verifiers do.

Method	Base models	Evaluation data	Factuality metric
Flame ( Lin et al., 2024 )	Llama-2 7B, 70B	Bio ( Min et al., 2023 ) , Alpaca-Fact ( Lin et al., 2024 ) , FAVA ( Mishra et al., 2024 )	FActScore (Inst-LLAMA ${+}$ npm)
		TruthfulQA ( Lin et al., 2022 ) , AlpacaEval ( Dubois et al., 2024 )	BLEU/ROUGE; GPT-4 win rate
FactAlign ( Huang and Chen, 2024 )	gemma-2b, Llama-3-8B , Phi3-Mini	LongFact-Objects ( Wei et al., 2024b )	SAFE F1@100 (ColBERT)
		FActScore biographies ( Min et al., 2023 )	FActScore (GPT-3.5 assessor)
KnowRL ( Ren et al., 2026 )	R1-Distill-Qwen-7B, Skywork-OR1-7B	SimpleQA ( Wei et al., 2024a ) , ChineseSimpleQA ( He et al., 2025 )	accuracy, non-rejected only
		TruthfulQA ( Lin et al., 2022 ) , GPQA ( Rein et al., 2024 ) , AIME	BLEU; accuracy

Appendix

Table 26: Base models, datasets, and metrics of the five methods. None shares a model or dataset with all others. Three metrics are precision-like, one credits refusals, and one excludes them.

Method	Training stages	Signal	What it optimizes
Flame	factuality-aware SFT $\to$ factuality-aware DPO	implicit	preference pairs scored with FActScore against retrieval
FactAlign	fKTO, from an instruction-tuned checkpoint that supplies the supervised stage	implicit	$\mathcal{L}_{\text{KTO}}+\lambda\,\mathcal{L}_{\text{fKTO}}$ : response-level KTO plus sentence-level labels
KnowRL	cold-start SFT $\to$ GRPO	rule	format $\pm 1$ , plus an LLM verdict $({+}2/{-}1/{+}1)$ , plus FActScore partial credit when wrong and not a refusal
FactGRPO	SFT (3k) $\to$ Dr. GRPO	rule	$-1$ if malformed, else $F/(T{+}1)+\lambda\log(1{+}F)+\mu w$ , with $\lambda_{\text{len}}{=}0$ , $\lambda_{\text{llm}}{=}0.1$ ; $S$ , $T$ from VeriScore
Faith	estimator $\to$ SFT $\to$ reward model $\to$ PPO $\to$ RAG rectifier (RAFT)	learned	a Llama-3-8B reward model with a scalar value head, trained to predict the EM_RP containment score

Appendix

Table 27: Training stages and optimization signals. Implicit rewards use

\beta\log(\pi/\pi_{\text{ref}})

; rule-based rewards use verifier outputs and checks; learned rewards use a trained model.

LLM Alignment

Large language model alignment studies methods for making models follow human intentions, values, and safety constraints across contexts.

See all 611 papers

Can We Trust the Judges? Validation of Factuality Evaluation Methods via Answer Perturbation
Sep 14, 2026Sarra Gharsallah, Adele Robaldo, Mariia Tokareva +5LLM Evaluation LLM-as-a-Judge
Data Science Department, EURECOM, Campus SophiaTech, Biot, France
FAITH: Factuality Alignment through Integrating Trustworthiness and Honestness
Apr 11, 2026Xiaoning Dong, Chengyan Wu, Yajie Wen +5LLM Alignment Factual Knowledge in Language Models
Tsinghua University, Institute for Interdisciplinary Information Sciences · Shanghai Qi Zhi Institute · South China Normal University, School of Electronic Science and Engineering +2
Beyond Precision: Importance-Aware Recall for Factuality Evaluation in Long-Form LLM Generation
Apr 3, 2026Nazanin Jafari, James Allan, Mohit IyyerLLM Evaluation Language Model Generation Evaluation
UMass Amherst · University of Maryland

Sep 14, 2026J/K move · Enter open · S save

Can We Trust the Judges? Validation of Factuality Evaluation Methods via Answer Perturbation

Sarra Gharsallah, Adele Robaldo, Mariia Tokareva, Giovanni Gatti Pinheiro+4

Data Science Department, EURECOM, Campus SophiaTech, Biot, France

LLM Evaluation LLM-as-a-Judge Factual Consistency Evaluation

Open paper AskPDF arXiv

Evaluating the factual correctness of large language models (LLMs) is vital for many applications. But are our evaluation tools themselves trustworthy? Despite the rise of factuality-based metrics, their sensitivity and reliability remain underexplored. This paper introduces a meta-evaluation framework that systematically tests these metrics using controlled corruptions of gold standard answers. Our method generates ranked outputs with known degrees of degradation to probe how metrics capture nuanced changes in truthfulness. Our experiments reveal that pipeline-based methods, such as the RAGAS's factual correctness metric, better track degradation than LLM-as-judge approaches. We also propose a new variant of the factual correctness metric that provides a competitive and cost-efficient.

Abstract

LLM Alignment

Explore similar work

Can We Trust the Judges? Validation of Factuality Evaluation Methods via Answer Perturbation

FAITH: Factuality Alignment through Integrating Trustworthiness and Honestness

Beyond Precision: Importance-Aware Recall for Factuality Evaluation in Long-Form LLM Generation