Papers › Harnessing the Power of Multi-Task Pretraining for Ground-Truth Level Natural Language...

Harnessing the Power of Multi-Task Pretraining for Ground-Truth Level Natural Language Explanations

8 Dec 2022arXiv:2212.04231archive 2025-07-28

Björn Plüster, Jakob Ambsdorf, Lukas Braach, Jae Hee Lee, Stefan Wermter

Natural language explanations promise to offer intuitively understandable explanations of a neural network's decision process in complex vision-language tasks, as pursued in recent VL-NLE models. While current models offer impressive performance on task accuracy and explanation plausibility, they suffer from a range of issues: Some models feature a modular design where the explanation generation module is poorly integrated with a separate module for task-answer prediction, employ backbone models trained on limited sets of tasks, or incorporate ad hoc solutions to increase performance on single datasets. We propose to evade these limitations by applying recent advances in large-scale multi-task pretraining of generative Transformer models to the problem of VL-NLE tasks. Our approach outperforms recent models by a large margin, with human annotators preferring the generated explanations over the ground truth in two out of three evaluated datasets. As a novel challenge in VL-NLE research, we propose the problem of multi-task VL-NLE and show that jointly training on multiple tasks can increase the explanation quality. We discuss the ethical implications of high-quality NLE generation and other issues in recent VL-NLE research.

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 8 code samples harvested from 1 repository linked to this paper: 7 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 1 honoured, 0 violated, 6 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code; here there are none.

By repository: official repository: 8 samples from 1 repository, 7 ran (7 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 between 2026-09-22 and 2026-09-26 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

ofa-x/ofa-x officialmentioned in papermentioned on GitHubpytorchApache-2.0Syntology: 8 samples, 7 ran (0 instrument-failed) · 1 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

8 samples harvested: 7 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 1 honoured, 0 violated, 6 with no contract checked; 0 where Syntology's instrument failed) · 1 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.

1ran · honoured contract
6ran
1unverified

Licence: 6 of the 8 samples are pointer only, meaning Syntology does not serve that copy's text. For 6 of them, this paper's copy is under a permissive licence and cleared, but the stored copy of the same code that Syntology serves is not; the licence label's tooltip gives Syntology's recorded reason. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from ofa-x/ofa-x. “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. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

ofa-x/ofa-x official repository 8 samples, 7 ran (0 instrument-failed) · 1 unverified
apply_half in 7 papers OFA/evaluate.py ran · honoured contract fingerprinted Apache-2.0 (permissive) · 357d3a4f0be679d9 · report fixture: signature t:Tensor/2/float · returned Tensor · level recorded 2026-09-22
construct_rdrop_sample OFA/criterions/label_smoothed_encouraging_loss.py ran Apache-2.0 (permissive) here; the stored copy is pointer only · e1397e09cb4c249d · report fixture: signature x:dict/0/? · returned dict · level recorded 2026-09-26
custom_to_pil in 9 papers OFA/criterions/clip_scst_loss.py ran Apache-2.0 (permissive) here; the stored copy is pointer only · 01478032e19f9e68 · report fixture: signature x:Tensor/3/float · returned Image · level recorded 2026-09-26
kl_loss in 37 papers
census: 8 implementations from 11 papers, 5 distinct outputs across 2 buckets, one shared input per bucket
OFA/criterions/label_smoothed_encouraging_loss.py ran fingerprinted Apache-2.0 (permissive) here; the stored copy is pointer only · e7939f8552fd8729 · report fixture: signature p:Tensor/2/float, q:Tensor/2/float · returned Tensor · level recorded 2026-09-26
label_smoothed_nll_loss in 43 papers OFA/criterions/label_smoothed_encouraging_loss.py ran Apache-2.0 (permissive) here; the stored copy is pointer only · cbd8615e3eba731b · report fixture: signature lprobs:Tensor/2/float, target:Tensor/1/int, epsilon:float/0/float, update_num:int/0/int, reduce:bool/0/bool, drop_worst_ratio:float/0/float, drop_worst_after:int/0/int, use_rdrop:bool/0/bool, reg_alpha:float/0/float, constraint_masks:NoneType/0/?, constraint_start:NoneType/0/?, constraint_end:NoneType/0/?, drop_best_ratio:float/0/float, drop_best_after:int/0/int · returned tuple · level recorded 2026-09-26
label_smoothed_nll_loss in 43 papers OFA/criterions/label_smoothed_cross_entropy.py ran Apache-2.0 (permissive) here; the stored copy is pointer only · cdbc222d0057fb31 · report fixture: signature lprobs:Tensor/2/float, target:Tensor/1/int, epsilon:float/0/float, update_num:int/0/int, reduce:bool/0/bool, drop_worst_ratio:float/0/float, drop_worst_after:int/0/int, use_rdrop:bool/0/bool, reg_alpha:float/0/float, constraint_masks:NoneType/0/?, constraint_start:NoneType/0/?, constraint_end:NoneType/0/? · returned tuple · level recorded 2026-09-26
label_smoothed_nll_loss in 43 papers OFA/criterions/label_smoothed_cross_entropy_expl.py ran Apache-2.0 (permissive) · 3a3448cb08438214 · report fixture: signature lprobs:Tensor/2/float, target:Tensor/1/int, epsilon:float/0/float, update_num:int/0/int, reduce:bool/0/bool, drop_worst_ratio:float/0/float, drop_worst_after:int/0/int, use_rdrop:bool/0/bool, reg_alpha:float/0/float, constraint_masks:NoneType/0/?, constraint_start:NoneType/0/?, constraint_end:NoneType/0/? · returned tuple · level recorded 2026-09-25
scst_loss OFA/criterions/clip_scst_loss.py unverified Apache-2.0 (permissive) here; the stored copy is pointer only · b0d5d75f27b8ce4f · report

Tasks

Explanation GenerationVisual EntailmentVisual Question Answering (VQA)

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Explanation Generation VCR OFA-X-MT Human Explanation Rating 77.3 #1 of 2 Archive leaderboard report
Explanation Generation VCR OFA-X Human Explanation Rating 68.9 #2 of 2 Archive leaderboard report
Explanation Generation VQA-X OFA-X Human Explanation Rating 89.5 #1 of 2 Archive leaderboard report
Explanation Generation VQA-X OFA-X-MT Human Explanation Rating 87.8 #2 of 2 Archive leaderboard report
Explanation Generation e-SNLI-VE OFA-X Human Explanation Rating 85.7 #1 of 2 Archive leaderboard report
Explanation Generation e-SNLI-VE OFA-X-MT Human Explanation Rating 80.4 #2 of 2 Archive leaderboard report
Visual Entailment e-SNLI-VE OFA-X Accuracy 80.9 #1 of 2 Archive leaderboard report
Visual Entailment e-SNLI-VE OFA-X-MT Accuracy 78.9 #2 of 2 Archive leaderboard report
Visual Question Answering (VQA) VCR (Q-A) test OFA-X Accuracy 71.2 #10 of 11 Archive leaderboard report
Visual Question Answering (VQA) VCR (Q-A) test OFA-X-MT Accuracy 62 #11 of 11 Archive leaderboard report
Visual Question Answering (VQA) VQA-X OFA-X-MT Accuracy 92.6 #1 of 2 Archive leaderboard report
Visual Question Answering (VQA) VQA-X OFA-X Accuracy 91.2 #2 of 2 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

Absolute Position EncodingsAdamAttentionBPEDense ConnectionsDropoutHOCLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-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