Papers › Sigmoid Loss for Language Image Pre-Training
Sigmoid Loss for Language Image Pre-Training
Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, Lucas Beyer
We propose a simple pairwise Sigmoid loss for Language-Image Pre-training (SigLIP). Unlike standard contrastive learning with softmax normalization, the sigmoid loss operates solely on image-text pairs and does not require a global view of the pairwise similarities for normalization. The sigmoid loss simultaneously allows further scaling up the batch size, while also performing better at smaller batch sizes. Combined with Locked-image Tuning, with only four TPUv4 chips, we train a SigLiT model that achieves 84.5% ImageNet zero-shot accuracy in two days. The disentanglement of the batch size from the loss further allows us to study the impact of examples vs pairs and negative to positive ratio. Finally, we push the batch size to the extreme, up to one million, and find that the benefits of growing batch size quickly diminish, with a more reasonable batch size of 32k being sufficient. We release our models at https://github.com/google-research/big_vision and hope our research motivates further explorations in improving the quality and efficiency of language-image pre-training.
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):
get_harvested_code_for_paper(arxiv_id="2303.15343")get_code(code_sha256="…")with a sample'scode_sha256from the list belowget_citation_path(paper_1="2303.15343", paper_2="…")with another paper's arXiv id or titleget_concepts_for_paper(arxiv_id="2303.15343")
Code
No sample from the archive's official repository (google-research/big_vision) is linked to this paper; the 27 samples below come from 6 community repositories, and 2 more are identical to code Syntology harvested from another paper's repository. Syntology harvested code from google-research/big_vision for 7 other papers.
google-research/big_vision: Syntology's graph links these samples to How to train your ViT? Data, Augmentation, and Regularization in Vision Transformers (arXiv:2106.10270; that paper's own run record: 2 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified), UViM: A Unified Modeling Approach for Vision with Learned Guiding Codes (arXiv:2205.10337; that paper's own run record: 2 ran (of which 1 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 1 where Syntology's instrument failed) · 1 unverified), PaLI: A Jointly-Scaled Multilingual Language-Image Model (arXiv:2209.06794; that paper's own run record: 2 ran (of which 1 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 1 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 2 unverified) and 4 more (every one in this page's JSON twin); nothing checks that they implement this paper's method.
Syntology 29 code samples: 27 from 6 repositories linked to this paper, 2 identical to code harvested from another paper's repository. Across all 29: 14 ran (of which 3 constructed an object rather than computing a result; 11 with no instrument failure: 0 honoured, 0 violated, 11 with no contract checked; 3 where Syntology's instrument failed) · 15 unverified. Where Syntology's instrument failed: 3 where our contract draft was wrong; failures of Syntology's instrument, not of the code.
By repository: community (archive-listed): 27 samples from 6 repositories, 14 ran (11 with no instrument failure; 3 where Syntology's instrument failed); 2 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 4 of its repositories that Syntology harvested for other papers:
- google-research/big_vision, official repository: harvested for 7 other papers, listed under the repository
- borisdayma/clip-jax, community (archive-listed): harvested for another paper, listed under the repository
- huggingface/transformers, community (archive-listed): harvested for 4 other papers, listed under the repository
- shkarupa-alex/tfclip, community (archive-listed): harvested for another paper, listed under the repository
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.
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
29 samples harvested: 14 ran (of which 3 constructed an object rather than computing a result; 11 with no instrument failure: 0 honoured, 0 violated, 11 with no contract checked; 3 where Syntology's instrument failed) · 15 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.
Licence: 25 of the 29 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 6 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. 2 samples are 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.
af515d1162a90440 · report
fixture: signature x:Tensor/4/float · returned Tensor · level recorded 2026-09-22 21e0abb626a7ae62 · report
fixture: signature x:Tensor/4/float · returned Tensor · level recorded 2026-09-22 c4c619fef7d59d34 · report
fixture: signature x:Tensor/3/float · returned Tensor · level recorded 2026-09-22 0748e821730b3fc1 · report
fixture: signature seq_len:int/0/int · returned Tensor · level recorded 2026-09-22 census: 9 implementations from 7 papers, 9 distinct outputs across 3 buckets, one shared input per bucket this sample: bucket 2 of 3, arg 1: rank 3, kind float, dtype float32 mobileclip/clip.py ran · metamorphic tier: invariant fingerprinted licence not identified · pointer only ·
88f8bcf4b76c2844 · report
fixture: signature x_q:Tensor/3/float, x_kv:NoneType/0/?, key_padding_mask:NoneType/0/?, attn_mask:NoneType/0/? · returned Tensor · level recorded 2026-09-22 census: 6 implementations from 6 papers, 4 distinct outputs across 3 buckets, one shared input per bucket mobileclip/clip.py ran · metamorphic tier: well formed licence not identified · pointer only ·
1e38402044297bef · report
fixture: signature seq_len:int/0/int · returned Tensor · level recorded 2026-09-22 59434b67e9a74429 · report
fixture: signature norm_type:str/0/str, num_features:int/0/int · returned LayerNormFP32 · level recorded 2026-09-22 a8fa6ab303249a21 · report
df9eaf41a5c09a07 · report
695f06cd58c0597e · report
43cafd406a2da3f3 · report
efcf8d9368ee9d08 · report
49fac99e1f1a0270 · report
88fa799b7ff64ccb · report
35094b7d325013c5 · report
eb62100582261eb4 · report
fixture: signature x:Tensor/2/float, x_prime:Tensor/2/float, tau:Tensor/0/float, b:Tensor/0/float, alpha:float/0/float, use_siglip:bool/0/bool, max_tau:float/0/float · returned tuple · level recorded 2026-09-22 67c4ef5004396e02 · report
fixture: signature logits:Tensor/2/float · returned Tensor · level recorded 2026-09-22 census: 5 implementations from 5 papers, 5 distinct outputs across 2 buckets, one shared input per bucket clip_siglip.py ran Apache-2.0 (permissive) ·
3586d486792b020d · report
fixture: signature output:list/1/dict · returned dict · level recorded 2026-09-25 ad49d2b1bad09561 · report
fixture: signature output:list/1/ndarray, labels:list/1/str · returned dict · level recorded 2026-09-24 997398ad737e7cd2 · report
fixture: returned NeighbourExchange · level recorded 2026-09-22 7caa2b64e6a6f021 · report
fixture: returned NeighbourExchangeBidir · level recorded 2026-09-22 2d2c2d56e476e772 · report
fixture: returned SigLipLoss · level recorded 2026-09-22 4a8006bab44bb658 · report
12c8728ab37fdce8 · report
2 samples are 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.
e332856e3c2fc814 · report
5b1fd364afcf3c05 · report
Method as code Syntology
What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it. Siblings are assemblies lifted from this paper in the same batch for the same method: alternative readings kept side by side; none replaces or supersedes another.
5 assemblies on this paper: 5 harvested, 0 generated, listed in 2 groups (siblings lifted in one batch form one group).
3 sibling assemblies for Sigmoid Loss (harvested), lifted in the same batch on 2026-09-09: alternative readings kept side by side; none replaces or supersedes another.
harvested Assembly for Sigmoid Loss (graph method: Sigmoid Loss). 1 brick; individually on a fixture: 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified. The assembly has not been run end-to-end.
one of 3 siblings
- siglip_loss entry · level 1 · ran · our draft was wrong ·
67c4ef5004396e02
harvested Assembly for Sigmoid Loss (graph method: Sigmoid Loss). 1 brick; individually on a fixture: 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified. The assembly has not been run end-to-end.
one of 3 siblings
- relic_loss entry · level 1 · ran · our draft was wrong ·
eb62100582261eb4
2 sibling assemblies for SigLIP (harvested), lifted in the same batch on 2026-09-09: alternative readings kept side by side; none replaces or supersedes another.
harvested Assembly for SigLIP (graph method: SigLIP). 8 bricks; individually on a fixture: 3 ran (of which 3 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 5 unverified; every one of the 3 samples that ran constructed an object rather than computing a result. The assembly has not been run end-to-end.
one of 2 siblings
- SigLipLoss entry · level 1 · ran ·
2d2c2d56e476e772 - _all_gather_with_grad helper · level 0 · unverified ·
9875e379a3b0cd8a - neighbour_exchange_bidir_with_grad helper · level 0 · unverified ·
4a8006bab44bb658 - neighbour_exchange_with_grad helper · level 0 · unverified ·
12c8728ab37fdce8 - NeighbourExchangeBidir helper · level 1 · ran ·
7caa2b64e6a6f021 - NeighbourExchange helper · level 1 · ran ·
997398ad737e7cd2 - neighbour_exchange_bidir helper · level 0 · unverified ·
5b1fd364afcf3c05 - neighbour_exchange helper · level 0 · unverified ·
e332856e3c2fc814
harvested Assembly for SigLIP (graph method: SigLIP). 16 bricks; individually on a fixture: 7 ran (of which 0 constructed an object rather than computing a result; 6 with no instrument failure: 0 honoured, 0 violated, 6 with no contract checked; 1 where Syntology's instrument failed) · 9 unverified. The assembly has not been run end-to-end.
one of 2 siblings
- CLIP entry · level 0 · unverified ·
a8fa6ab303249a21 - MCi helper · level 0 · unverified ·
579a47ae5bfe8248 - TextTransformer helper · level 0 · unverified ·
88fa799b7ff64ccb - GlobalPool2D helper · level 1 · ran ·
21e0abb626a7ae62 - PositionalEmbedding helper · level 1 · ran ·
1e38402044297bef - RepMixerBlock helper · level 0 · unverified ·
efcf8d9368ee9d08 - TransformerEncoder helper · level 0 · unverified ·
35094b7d325013c5 - get_normalization_layer helper · level 1 · ran · our draft was wrong ·
59434b67e9a74429 - GlobalPool helper · level 1 · ran ·
af515d1162a90440 - LearnablePositionalEmbedding helper · level 1 · ran ·
0748e821730b3fc1 - ConvFFN helper · level 0 · unverified ·
df9eaf41a5c09a07 - RepMixer helper · level 0 · unverified ·
43cafd406a2da3f3 - MultiHeadAttention helper · level 1 · ran ·
88f8bcf4b76c2844 - LayerNormFP32 helper · level 1 · ran ·
c4c619fef7d59d34 - MobileOneBlock helper · level 0 · unverified ·
695f06cd58c0597e - SEBlock helper · level 0 · unverified ·
49fac99e1f1a0270
Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2303.15343") returns these assemblies with each brick's role and wiring (how to connect).
Tasks
Results from the paper archive 2025-07-28
| Task | Dataset | Model | Metric | Value | Rank at snapshot | Leaderboard | Report |
|---|---|---|---|---|---|---|---|
| Image-to-Text Retrieval | COCO | SigLIP (ViT-L, zero-shot) | Recall@1 | 70.6 | #1 of 1 | 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
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