Papers › Dynamic Routing Between Capsules
Dynamic Routing Between Capsules
Sara Sabour, Nicholas Frosst, Geoffrey E. Hinton
A capsule is a group of neurons whose activity vector represents the instantiation parameters of a specific type of entity such as an object or an object part. We use the length of the activity vector to represent the probability that the entity exists and its orientation to represent the instantiation parameters. Active capsules at one level make predictions, via transformation matrices, for the instantiation parameters of higher-level capsules. When multiple predictions agree, a higher level capsule becomes active. We show that a discrimininatively trained, multi-layer capsule system achieves state-of-the-art performance on MNIST and is considerably better than a convolutional net at recognizing highly overlapping digits. To achieve these results we use an iterative routing-by-agreement mechanism: A lower-level capsule prefers to send its output to higher level capsules whose activity vectors have a big scalar product with the prediction coming from the lower-level capsule.
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="1710.09829")get_code(code_sha256="…")with a sample'scode_sha256from the list belowget_citation_path(paper_1="1710.09829", paper_2="…")with another paper's arXiv id or titleget_concepts_for_paper(arxiv_id="1710.09829")
Code
Syntology 118 code samples harvested from 36 repositories linked to this paper: 75 ran (of which 0 constructed an object rather than computing a result; 58 with no instrument failure: 0 honoured, 0 violated, 58 with no contract checked; 17 where Syntology's instrument failed) · 43 unverified. Where Syntology's instrument failed: 15 where our contract draft was wrong; 2 where the contract check could not run; failures of Syntology's instrument, not of the code.
By repository: official repository: 2 samples from 1 repository, 1 ran (1 with no instrument failure; 0 where Syntology's instrument failed); community (archive-listed): 116 samples from 35 repositories, 74 ran (57 with no instrument failure; 17 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.
No sample is linked to this paper from 2 of its repositories that Syntology harvested for other papers:
- hli2020/nn_capsulation, community (archive-listed): harvested for another paper, listed under the repository
- labmlai/annotated_deep_learning_paper_implementations, community (archive-listed): harvested for 15 other papers, listed under the repository
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.
77 repositories listed; official and paper-mentioned ones first.
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
118 samples harvested: 75 ran (of which 0 constructed an object rather than computing a result; 58 with no instrument failure: 0 honoured, 0 violated, 58 with no contract checked; 17 where Syntology's instrument failed) · 43 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: 13 of the 118 samples are pointer only, meaning Syntology does not serve that copy's text. For 1 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 36 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. 1 sample whose link to this paper records no repository is listed under one of this paper's own repositories, because identical code was harvested from it for another paper. “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.
9752e0af7e2e920e · report
fixture: signature is_training:bool/0/bool · returned dict · level recorded 2026-09-25 b1b870e26b8c5598 · report
fa45f6e21e1f5d6e · report
fixture: signature x:ndarray/3/int, axis:int/0/int · returned ndarray · level recorded 2026-09-22 c5a569bffcb9d305 · report
fixture: signature image:ndarray/3/float, salt:float/0/float, amount:float/0/float · returned ndarray · level recorded 2026-09-22 0fa7ba2d6f1b5e7d · report
fixture: signature folder:str/0/str · returned ndarray · level recorded 2026-09-26 9e266efd2f9b3537 · report
ef6f7f989e63e349 · report
census: 8 implementations from 5 papers, 4 distinct outputs across 2 buckets, one shared input per bucket this sample: bucket 1 of 2, arg 1: rank 2, kind float, dtype float32 CapsDiscriminator.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 fingerprinted licence of this copy not recorded ·
e8e375ca46107111 · report
fixture: signature inputs:Tensor/2/float, axis:int/0/int · returned Tensor · level recorded 2026-09-22 6d94bff65a753b0d · report
fixture: signature y_true:Tensor/2/float, y_pred:Tensor/2/float, x:Tensor/4/float, x_recon:Tensor/4/float, lam_recon:float/0/float · returned Tensor · level recorded 2026-09-22 f854823338dc9ee8 · report
a5be0ded65ff286c · report
643e4ac03170a791 · report
fixture: signature masked_input:EagerTensor/2/float, target_shape:tuple/1/int, dense_units:list/1/int, output_units:int/0/int, activation:str/0/str · returned EagerTensor · level recorded 2026-09-26 census: 8 implementations from 5 papers, 4 distinct outputs across 2 buckets, one shared input per bucket src/capsule_networks.py ran MIT (permissive) ·
fed6154c45168494 · report
fixture: signature vector:EagerTensor/3/float, ax:int/0/int, epsilon:float/0/float · returned EagerTensor · level recorded 2026-09-26 b72eda3284443993 · report
fixture: signature data_dir:str/0/str · returned tuple · level recorded 2026-09-22 e681684c378bcb1e · report
fixture: signature x:Tensor/1/float · returned Tensor · level recorded 2026-09-26 45ba4b2229e7c5d0 · report
fixture: signature image_string:bytes/0/?, classes:list/1/int · returned Example · level recorded 2026-09-26 d47f898b34da72b5 · report
fixture: signature x:EagerTensor/4/float · returned SymbolicTensor · level recorded 2026-09-26 ad88fd8ea70fc685 · report
fixture: signature x:EagerTensor/4/float · returned EagerTensor · level recorded 2026-09-26 59858e235d46459f · report
fixture: signature img:ndarray/3/float, shape:list/1/int · returned ndarray · level recorded 2026-09-26 0da33f8508f5c3ef · report
fixture: signature img:ndarray/3/int, shape:int/0/int · returned ndarray · level recorded 2026-09-26 454894a6d26864c2 · report
fixture: signature path:str/0/str, mode:str/0/str · returned ndarray · level recorded 2026-09-26 904b93c8951de375 · report
78eab6842a55500d · report
fixture: signature folder:str/0/str · returned tuple · level recorded 2026-09-26 1e8dccecc001d2fd · report
fixture: signature train_list_path:str/0/str, batch_size:int/0/int · returned function · level recorded 2026-09-26 census: 8 implementations from 5 papers, 4 distinct outputs across 2 buckets, one shared input per bucket this sample: bucket 1 of 2, arg 1: rank 2, kind float, dtype float32 capsnet.py ran · our draft was wrong fingerprinted MIT (permissive) ·
89d981e0b618190e · report
fixture: signature x:Tensor/2/float, dim:int/0/int · returned Tensor · level recorded 2026-09-22 22def68ffad5b584 · report
census: 53 implementations from 63 papers, 16 distinct outputs across 8 buckets, one shared input per bucket src/digitcaps.py ran fingerprinted MIT (permissive) ·
2719723f2846d85e · report
fixture: signature input_tensor:Tensor/4/float, dim:int/0/int · returned Tensor · level recorded 2026-09-26 a4f9b362c3fb2a02 · report
fixture: signature generated_images:ndarray/4/float, height:NoneType/0/?, width:NoneType/0/? · returned ndarray · level recorded 2026-09-26 census: 8 implementations from 5 papers, 4 distinct outputs across 2 buckets, one shared input per bucket capsulelayers.py unverified MIT (permissive) ·
bc53226908b5d193 · report
census: 8 implementations from 5 papers, 4 distinct outputs across 2 buckets, one shared input per bucket this sample: bucket 2 of 2, arg 1: rank 3, kind float, dtype float32 net.py ran · our draft was wrong fingerprinted BSD-3-Clause (permissive) ·
5abcd8c6705f2368 · report
fixture: signature x:Tensor/3/float · returned Tensor · level recorded 2026-09-22 2e8edc862c5c964a · report
fixture: signature img:ndarray/2/int, mask:ndarray/2/int, grid_size:int/0/int · returned tuple · level recorded 2026-09-26 f47fb66b0f1495bb · report
fixture: signature tup:tuple/1/ndarray · returned ndarray · level recorded 2026-09-26 ba9a20e6fac96f58 · report
fixture: signature my_list:list/1/ndarray · returned ndarray · level recorded 2026-09-26 e161978d5621fd5a · report
fixture: signature labels_dense:ndarray/1/int, num_classes:int/0/int · returned ndarray · level recorded 2026-09-24 a4fd5ddd12015349 · report
fixture: signature f:BytesIO/0/?, one_hot:bool/0/bool, num_classes:int/0/int · returned ndarray · level recorded 2026-09-26 cd1f1b152fc81869 · report
fixture: signature img:ndarray/2/int, centers:ndarray/2/int, patch_size:int/0/int, n_patches:int/0/int, augment:bool/0/bool · returned ndarray · level recorded 2026-09-26 9609ff43029770e8 · report
fixture: signature img:ndarray/3/int, centers:ndarray/2/int, augment:bool/0/bool · returned ndarray · level recorded 2026-09-26 2ecfcd586e6b8acf · report
fixture: signature centers:ndarray/2/float · returned ndarray · level recorded 2026-09-26 3619549fb958505d · report
fixture: signature I:ndarray/3/int · returned tuple · level recorded 2026-09-26 c2c66bc5a1670838 · report
fixture: signature I:ndarray/3/int · returned tuple · level recorded 2026-09-26 7e43778f815ea3c2 · report
fixture: signature I1:ndarray/2/float, I2:ndarray/2/float, I3:ndarray/2/float · returned ndarray · level recorded 2026-09-26 7b42b0ac5492beb9 · report
fixture: signature img:ndarray/3/int, fmask:ndarray/2/int, all_masks:ndarray/2/int, clipLimit:float/0/float · returned tuple · level recorded 2026-09-26 c95802bf73f69647 · report
fixture: signature path:str/0/str · returned ndarray · level recorded 2026-09-26 5e53f8244d86e2f9 · report
fixture: signature output_vectors:ndarray/5/float, d_caps2:int/0/int, min:float/0/float, max:float/0/float, n_steps:int/0/int · returned ndarray · level recorded 2026-09-26 census: 8 implementations from 5 papers, 4 distinct outputs across 2 buckets, one shared input per bucket networks/capsnet.py unverified MIT (permissive) ·
42ce6a693b03e2f2 · report
5568e495aa546957 · report
ae560fa01740cfe8 · report
fixture: signature input_dim:int/0/int · returned Sequential · level recorded 2026-09-26 f57a3bb8459d9735 · report
fixture: signature prefix:str/0/str · returned tuple · level recorded 2026-09-26 93341244192ac2a3 · report
fixture: signature path:str/0/str · returned tuple · level recorded 2026-09-26 09b60348edfeca3d · report
fixture: signature sequences:list/1/list, dimension:int/0/int · returned ndarray · level recorded 2026-09-26 e9572b0ab2492828 · report
census: 8 implementations from 5 papers, 4 distinct outputs across 2 buckets, one shared input per bucket this sample: bucket 1 of 2, arg 1: rank 2, kind float, dtype float32 capsules.py ran · our draft was wrong fingerprinted MIT (permissive) ·
36782ecd204f05d4 · report
fixture: signature s:Tensor/2/float, dim:int/0/int · returned Tensor · level recorded 2026-09-22 0d412a9be3a2a5d6 · report
fixture: signature signal:ndarray/2/float, snr:float/0/float · returned ndarray · level recorded 2026-09-26 958171517f243d79 · report
fixture: signature y_pred:Tensor/2/float, y_true:Tensor/2/float, coefficient:float/0/float · returned Tensor · level recorded 2026-09-26 census: 8 implementations from 5 papers, 4 distinct outputs across 2 buckets, one shared input per bucket this sample: bucket 2 of 2, arg 1: rank 3, kind float, dtype float32 functional.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) ·
43a28458e2019a51 · report
fixture: signature input:Tensor/3/float, dim:int/0/int, keepdim:bool/0/bool · returned Tensor · level recorded 2026-09-22 ef26365c2b293ba7 · report
fixture: signature batch_size:int/0/int, valid_size:float/0/float, train_dataset:SimpleDataset/0/?, valid_dataset:SimpleDataset/0/?, test_dataset:SimpleDataset/0/? · returned tuple · level recorded 2026-09-26 census: 9 implementations from 11 papers, 8 distinct outputs across 4 buckets, one shared input per bucket tools.py ran fingerprinted MIT (permissive) ·
5c5b8a89cd91d2f6 · report
fixture: signature image:Tensor/2/float · returned Tensor · level recorded 2026-09-26 5c2f7dad1037f39e · report
fixture: signature SAVE_DIR:str/0/str, filename:str/0/str · returned str · level recorded 2026-09-26 census: 8 implementations from 5 papers, 4 distinct outputs across 2 buckets, one shared input per bucket tools.py ran fingerprinted MIT (permissive) ·
e6c4c09caf295822 · report
fixture: signature x:Tensor/2/float, dim:int/0/int · returned Tensor · level recorded 2026-09-26 916d6f93349e426b · report
fixture: signature x:Tensor/4/float, max_shift:int/0/int · returned Tensor · level recorded 2026-09-22 census: 53 implementations from 63 papers, 16 distinct outputs across 8 buckets, one shared input per bucket this sample: bucket 7 of 8, arg 1: rank 3, kind float, dtype float32 capsule_network.py ran · our draft was wrong fingerprinted no licence file found · pointer only ·
d55eaf6b69be6376 · report
fixture: signature input:Tensor/3/float, dim:int/0/int · returned Tensor · level recorded 2026-09-22 9169bb6c16583f82 · report
fixture: signature shapes:tuple/1/tuple · returned tuple · level recorded 2026-09-22 c4647dd5ef46a3fd · report
fixture: signature input:Tensor/3/float, factor:float/0/float · returned Tensor · level recorded 2026-09-22 census: 8 implementations from 5 papers, 4 distinct outputs across 2 buckets, one shared input per bucket this sample: bucket 2 of 2, arg 1: rank 3, kind float, dtype float32 layers/cap_layer.py ran · our draft was wrong no licence file found · pointer only ·
ac7650b0121cd6c9 · report
fixture: signature vec:Tensor/3/float, manner:str/0/str · returned Tensor · level recorded 2026-09-22 c0f267f39366e309 · report
fixture: signature rc_cap:Tensor/2/float · returned Tensor · level recorded 2026-09-22 6c522a777d335f98 · report
fixture: signature input_shape:tuple/1/int · returned Functional · level recorded 2026-09-26 9e374d498da47a80 · report
fixture: signature result:ndarray/1/int, reference:ndarray/1/int · returned float64 · level recorded 2026-09-26 4d42218ea782618b · report
fixture: signature y_true:EagerTensor/4/float, y_pred:EagerTensor/4/float, threshold:float/0/float, axis:list/1/int, smooth:float/0/float · returned EagerTensor · level recorded 2026-09-26 354cb4ef853ea7c0 · report
fixture: signature image:ndarray/3/float, alpha:int/0/int, sigma:int/0/int, alpha_affine:int/0/int, random_state:RandomState/0/? · returned ndarray · level recorded 2026-09-26 bebfab67566d4deb · report
fixture: signature result:ndarray/2/int, reference:ndarray/2/int · returned float · level recorded 2026-09-26 7f6b9d28c90431fb · report
fixture: signature result:ndarray/1/int, reference:ndarray/1/int · returned float64 · level recorded 2026-09-26 df095bccc89495e4 · report
fixture: signature x:ndarray/4/float, data_format:NoneType/0/? · returned ndarray · level recorded 2026-09-26 fe870e6febcb38d2 · report
census: 10 implementations from 16 papers, 7 distinct outputs across 4 buckets, one shared input per bucket custom_losses.py unverified Apache-2.0 (permissive) ·
89675961d6c5df8a · report
7b208df6adf33efe · report
25c21be9120353e4 · report
9baf4429c73eff71 · report
fixture: signature model:callable/0/?, data:tuple/1/Tensor · returned DummyStatistics · level recorded 2026-09-26 census: 8 implementations from 5 papers, 4 distinct outputs across 2 buckets, one shared input per bucket capsulelayers.py unverified MIT (permissive) ·
d1e743564e5d4d8c · report
889c715d11d8841d · report
fixture: signature generated_images:ndarray/4/int · returned ndarray · level recorded 2026-09-26 8278b9f74e0d99c0 · report
fixture: signature labels:EagerTensor/2/float, logits:EagerTensor/2/float, upper_margin:float/0/float, bottom_margin:float/0/float, downweight:float/0/float · returned EagerTensor · level recorded 2026-09-26 71f308af9ead46b8 · report
fixture: signature a:EagerTensor/2/float, b:EagerTensor/2/float, transpose_a:bool/0/bool, transpose_b:bool/0/bool, name:NoneType/0/? · returned SymbolicTensor · level recorded 2026-09-26 94a9d3fa58aca150 · report
fixture: signature inputs:EagerTensor/2/float, name:NoneType/0/? · returned list · level recorded 2026-09-26 census: 6 implementations from 8 papers, 5 distinct outputs across 2 buckets, one shared input per bucket capslayer/ops/losses/losses.py unverified Apache-2.0 (permissive) ·
512bf4bce2286e2e · report
25474dc6ec455a0d · report
fixture: signature tower_grads:list/1/list · returned list · level recorded 2026-09-26 a60140719ee8eeab · report
fixture: signature batch_size:int/0/int, is_training:bool/0/bool · returned tuple · level recorded 2026-09-26 census: 6 implementations from 8 papers, 5 distinct outputs across 2 buckets, one shared input per bucket capslayer/ops/losses/losses.py ran Apache-2.0 (permissive) ·
aeb7fd30067180af · report
fixture: signature labels:EagerTensor/1/int, logits:EagerTensor/2/float, regularizer:EagerTensor/1/float · returned SymbolicTensor · level recorded 2026-09-26 census: 8 implementations from 7 papers, 7 distinct outputs across 6 buckets, one shared input per bucket capsule_network.py ran · our draft was wrong MIT (permissive) ·
64858477ea4e51d9 · report
fixture: signature y:Tensor/2/float, pred:Tensor/1/int · returned float64 · level recorded 2026-09-22 19f6c412420d3c16 · report
fixture: signature dataset:StandInDataset/0/? · returned list · level recorded 2026-09-22 d805a7b2c2d7d3ad · report
fixture: signature dataset_name:str/0/str · returned ndarray · level recorded 2026-09-26 18f1ac5a3bb1245f · report
fixture: signature dataset_name:str/0/str · returned int · level recorded 2026-09-26 4293fbb258948def · report
fixture: signature dataset_name:str/0/str · returned int · level recorded 2026-09-26 c23def11ceb6917f · report
fixture: signature input:EagerTensor/4/float, kernel:int/0/int, stride:int/0/int · returned SymbolicTensor · level recorded 2026-09-26 Tasks
Datasets
Introduced by this paper, per the archive.
Results from the paper archive 2025-07-28
| Task | Dataset | Model | Metric | Value | Rank at snapshot | Leaderboard | Report |
|---|---|---|---|---|---|---|---|
| Image Classification | CIFAR-10 | ensemble of 7 models | Percentage correct | 89.4 | #209 of 265 | Archive leaderboard | report |
| Image Classification | EMNIST-Balanced | TextCaps | Accuracy | 90.46 | #6 of 20 | Archive leaderboard | report |
| Image Classification | MNIST | CapsNet | Percentage error | 0.25 | #12 of 81 | Archive leaderboard | report |
| Image Classification | MultiMNIST | CapsNet | Percentage error | 5.2 | #1 of 1 | Archive leaderboard | report |
| Image Classification | smallNORB | CapsNet | Classification Error | 3.77 | #6 of 7 | 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
Introduced by this paper: CapsNet, Capsule Network
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