Papers › Self-critical Sequence Training for Image Captioning

Self-critical Sequence Training for Image Captioning

2 Dec 2016CVPR 2017 7arXiv:1612.00563archive 2025-07-28

Steven J. Rennie, Etienne Marcheret, Youssef Mroueh, Jarret Ross, Vaibhava Goel

Recently it has been shown that policy-gradient methods for reinforcement learning can be utilized to train deep end-to-end systems directly on non-differentiable metrics for the task at hand. In this paper we consider the problem of optimizing image captioning systems using reinforcement learning, and show that by carefully optimizing our systems using the test metrics of the MSCOCO task, significant gains in performance can be realized. Our systems are built using a new optimization approach that we call self-critical sequence training (SCST). SCST is a form of the popular REINFORCE algorithm that, rather than estimating a "baseline" to normalize the rewards and reduce variance, utilizes the output of its own test-time inference algorithm to normalize the rewards it experiences. Using this approach, estimating the reward signal (as actor-critic methods must do) and estimating normalization (as REINFORCE algorithms typically do) is avoided, while at the same time harmonizing the model with respect to its test-time inference procedure. Empirically we find that directly optimizing the CIDEr metric with SCST and greedy decoding at test-time is highly effective. Our results on the MSCOCO evaluation sever establish a new state-of-the-art on the task, improving the best result in terms of CIDEr from 104.9 to 114.7.

PaperPDFConference PDFCodeCode 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 13 code samples: 12 from 3 repositories linked to this paper, 1 identical to code harvested from another paper's repository. Across all 13: 12 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 0 honoured, 1 violated, 4 with no contract checked; 7 where Syntology's instrument failed) · 1 unverified. Where Syntology's instrument failed: 7 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: community (archive-listed): 12 samples from 3 repositories, 11 ran (5 with no instrument failure; 6 where Syntology's instrument failed); 1 identical to code first harvested elsewhere. The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

No sample is linked to this paper from 5 of its repositories that Syntology harvested for other papers:

Levels written between 2026-09-22 and 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed. 6 of the 12 samples that ran have no recorded date.

31 repositories listed; official and paper-mentioned ones first.

Dlut-lab-zmn/Image-Captioning-Attack mentioned on GitHubpytorchSyntology: not harvested report
FJSam/SelfCritical_ImageCaptioning mentioned on GitHubpytorchSyntology: not harvested report
Kumar-Tarun/summarization mentioned on GitHubpytorchSyntology: not harvested report
SJTU-VLG/ImageCaption-LBPF-qy mentioned on GitHubpytorchSyntology: not harvested report
Zhendong-Wang/arsm_image_captioning mentioned on GitHubpytorchSyntology: not harvested report
arjung128/image-paragraph-captioning mentioned on GitHubpytorchSyntology: not harvested report
hjjpku/dynamic_graph_caption mentioned on GitHubpytorchSyntology: not harvested report
husthuaan/AAT mentioned on GitHubpytorchMITSyntology: 8 samples, 7 ran (3 instrument-failed) · 1 unverified report
husthuaan/AoANet mentioned on GitHubpytorchMITSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Attention on Attention for Image Captioning (arXiv:1908.06954; that paper's own run record: 9 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 0 honoured, 0 violated, 5 with no contract checked; 4 where Syntology's instrument failed) · 5 unverified); nothing checks that they implement this paper's method. report
jchenghu/sacreeos mentioned on GitHubpytorchApache-2.0Syntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to A Call for Clarity in Reporting BLEU Scores (arXiv:1804.08771; that paper's own run record: 11 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 2 honoured, 0 violated, 7 with no contract checked; 2 where Syntology's instrument failed) · 0 unverified); nothing checks that they implement this paper's method. report
jiahuei/COMIC-Compact-Image-Captioning-with-Attention mentioned on GitHubtfSyntology: not harvested report
jiahuei/Self-Critical-SCST-TensorFlow mentioned on GitHubtfSyntology: not harvested report
kangkang59812/GraphCaption mentioned on GitHubpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Bottom-Up and Top-Down Attention for Image Captioning and Visual Question Answering (arXiv:1707.07998; that paper's own run record: 9 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 1 violated, 0 with no contract checked; 8 where Syntology's instrument failed) · 0 unverified); nothing checks that they implement this paper's method. report
king-zark/self_critical mentioned on GitHubpytorchSyntology: not harvested report
krasserm/fairseq-image-captioning mentioned on GitHubpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Show, Attend and Tell: Neural Image Caption Generation with Visual Attention (arXiv:1502.03044; that paper's own run record: 58 ran (of which 0 constructed an object rather than computing a result; 45 with no instrument failure: 3 honoured, 0 violated, 42 with no contract checked; 13 where Syntology's instrument failed) · 26 unverified); nothing checks that they implement this paper's method. report
lukemelas/image-paragraph-captioning mentioned on GitHubpytorchSyntology: not harvested report
mazm13/Image-to-Tree.pytorch mentioned on GitHubpytorchSyntology: not harvested report
mykolesiko/nlp_projects mentioned on GitHubpytorchSyntology: not harvested report
quantum-fusion/AI-deeplearning mentioned on GitHubpytorchSyntology: not harvested report
ruotianluo/GoogleConceptualCaptioning mentioned on GitHubpytorchSyntology: 2 samples, 2 ran (2 instrument-failed) · 0 unverified report
ruotianluo/ImageCaptioning.pytorch mentioned on GitHubpytorchSyntology: not harvested report
ruotianluo/Transformer_Captioning mentioned on GitHubpytorchSyntology: 2 samples, 2 ran (1 instrument-failed) · 0 unverified report
ruotianluo/neuraltalk2.pytorch mentioned on GitHubpytorchSyntology: not harvested report
ruotianluo/self-critical.pytorch mentioned on GitHubpytorchSyntology: not harvested report
s-52/ImageCaptions mentioned on GitHubpytorchSyntology: not harvested report
sgondala/GoogleConceptualCaptioning mentioned on GitHubpytorchSyntology: not harvested report
ukyh/switch_disc_caption mentioned on GitHubpytorchSyntology: not harvested report
wtliao/ImageTransformer mentioned on GitHubpytorchSyntology: not harvested report
xiaobai714/image_caption mentioned on GitHubpytorchSyntology: not harvested report
ylqi/gl-rg mentioned on GitHubpytorchMITSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to GL-RG: Global-Local Representation Granularity for Video Captioning (arXiv:2205.10706; that paper's own run record: 2 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 0 where Syntology's instrument failed) · 3 unverified); nothing checks that they implement this paper's method. 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.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

13 samples harvested: 12 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 0 honoured, 1 violated, 4 with no contract checked; 7 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 · violated contract
7ran · our draft was wrong
4ran
1unverified

Licence: 3 of the 13 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 3 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. 2 samples whose link to this paper records no repository are listed under one of this paper's own repositories, because identical code was harvested from it for another paper. 1 sample is listed last, in a group without a repository. “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.

husthuaan/AAT community (archive-listed) 8 samples, 7 ran (3 instrument-failed) · 1 unverified
count_bad in 5 papers eval_utils.py ran fingerprinted MIT (permissive) · 0a4de8f498e3e70c · report fixture: signature sen:str/0/str · returned int · level recorded 2026-09-25
pack_wrapper in 13 papers models/AttModel.py ran · our draft was wrong MIT (permissive) · d2379d710eedc4ae · report fixture: signature module:callable/0/?, att_feats:Tensor/3/float, att_masks:Tensor/2/float · returned Tensor · level recorded, date not recorded
pad_unsort_packed_sequence in 13 papers models/AttModel.py ran · our draft was wrong MIT (permissive) · bfac58a04b6835f5 · report fixture: signature input:PackedSequence/1/Tensor, inv_ix:Tensor/1/int · returned Tensor · level recorded, date not recorded
resnet18 in 432 papers misc/resnet.py ran MIT (permissive) · 2b17380e9938d559 · report fixture: signature pretrained:bool/0/bool · returned ResNet · level recorded 2026-09-25
resnet34 in 148 papers misc/resnet.py ran MIT (permissive) · d206d964368eeb24 · report fixture: signature pretrained:bool/0/bool · returned ResNet · level recorded 2026-09-25
resnet50 in 121 papers misc/resnet.py ran MIT (permissive) · fade30a7dab7a039 · report fixture: signature pretrained:bool/0/bool · returned ResNet · level recorded 2026-09-25
sort_pack_padded_sequence in 13 papers models/AttModel.py ran · our draft was wrong fingerprinted MIT (permissive) · e56d2f9cbd9cb8b1 · report fixture: signature input:Tensor/3/float, lengths:Tensor/1/int · returned tuple · level recorded, date not recorded
language_eval in 7 papers eval_utils.py unverified MIT (permissive) · 06b72839dcd7433f · report
ruotianluo/GoogleConceptualCaptioning community (archive-listed) 2 samples, 2 ran (2 instrument-failed) · 0 unverified
build_vocab in 36 papers scripts/prepro_labels.py ran · our draft was wrong MIT (permissive) · 41b9ec0c269d8a91 · report fixture: signature imgs:list/1/dict, params:dict/0/? · returned list · level recorded 2026-09-22
encode_captions scripts/prepro_labels.py ran · our draft was wrong MIT (permissive) · 4aad8242843ac47b · report fixture: signature imgs:list/1/dict, params:dict/0/?, wtoi:dict/0/? · returned tuple · level recorded 2026-09-22
ruotianluo/Transformer_Captioning community (archive-listed) 2 samples, 2 ran (1 instrument-failed) · 0 unverified
attention in 144 papers
census: 11 implementations from 10 papers, 7 distinct outputs across 5 buckets, one shared input per bucket
models/TransformerModel.py identical code Syntology harvested from this repository for another paper; this paper's link records no repository, path or licence of its own ran · our draft was wrong licence of this copy not recorded · 20bb0ff4a2d77a03 · report fixture: signature query:Tensor/3/float, key:Tensor/3/float, value:Tensor/3/float, mask:Tensor/4/float, dropout:callable/0/? · returned tuple · level recorded, date not recorded
subsequent_mask in 38 papers models/TransformerModel.py identical code Syntology harvested from this repository for another paper; this paper's link records no repository, path or licence of its own ran · violated contract licence of this copy not recorded · dd03765c7da9ca51 · report fixture: signature size:int/0/int · returned Tensor · level recorded, date not recorded
Repository not located 1 sample, 1 ran (1 instrument-failed) · 0 unverified
1 sample is identical to code Syntology harvested from another paper's repository; this paper's own copy was not located, so no repository is named and no licence is recorded.
clones in 92 papers repository not located ran · our draft was wrong licence of this copy not recorded · a3722169bbc81569 · report fixture: signature module:callable/0/?, N:int/0/int · returned ModuleList · level recorded, date not recorded

Tasks

Image CaptioningPolicy Gradient MethodsReinforcement LearningReinforcement Learning (RL)reinforcement-learning

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

Introduced by this paper: SCST

REINFORCESCST

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