Home › Census › timestep_embedding

timestep_embedding

Syntologyfunction-name censuscensus 2026-09-22battery b986f7e04d79all samples with this name

timestep_embedding: 22 implementations from 27 papers ran on one shared input (census 2026-09-22, battery b986f7e04d79); they produced 6 distinct outputs across 2 buckets, one shared input per bucket.

Identical values to six decimals (the recorded digest) on the shared input are agreement on those inputs, not equivalence. Implementations are compared only within one bucket, the positional (rank, kind, dtype) of each array argument; the argument name is not part of the key because the harness draws the shared array from (rank, kind) and casts it to the dtype, whatever the name; each bucket's shared input is fixed by that key, so members of one bucket saw bitwise-identical inputs under their own scalar arguments. A cluster is the set of members whose recorded output digest is identical. Nothing here says which computation a paper's method intended, and nothing reproduces a paper's results.

Not compared, and not in the tables or the counts above:

Bucket 1 of 2: arg 1: rank 1, kind float, dtype float32

13 implementations from 17 papers share this bucket (rank, kind, dtype of each array argument, positional; each member's recorded signature, argument name included, is shown under it); 3 distinct outputs, largest cluster first. Values are the first 8 of the recorded output, flattened.

Cluster (same digest to six decimals)MembersShared output on this bucket's input
10 implementations
13 papers
cddd37cd7c51
recorded values identical
one code sha held from 3 papers' repositories
libs/uvit.py 262b18d8
recorded timesteps:1/float/float32; dim=4, max_period=10000
one code sha held from 2 papers' repositories
see2sound/codi/models/latent_diffusion/diffusion_unet.py 7e83c65f
recorded timesteps:1/float/float32; dim=4, max_period=10000, repeat_only=False

hyvideo/modules/models.py 1919bf55
recorded t:1/float/float32; dim=4, max_period=10000

models/unet.py 1b261a79
recorded timesteps:1/float/float32; dim=4, max_period=10000, repeat_only=False

model/BrownianBridge/BrownianBridgeModel_c2v.py 1d718716
recorded timesteps:1/float/float32; dim=4, max_period=10000, repeat_only=False

model/RecModel.py 59afc50c
recorded timesteps:1/float/float32; dim=4, device='cpu', max_period=10000

JGA-LBD/DDBM/ddbm/unet3d.py 6b0b644d
recorded timesteps:1/float/float32; dim=4, max_period=10000, dtype=torch.float32

model/big_unet.py b0aa348a
recorded timesteps:1/float/float32; dim=4, max_period=10000

model/networks/diffusion_networks/sg_diff.py b4f50214
recorded timesteps:1/float/float32; dim=4, max_period=10000, repeat_only=False

models/amfd_loss.py ecd79b71
recorded t:1/float/float32; dim=4, max_period=10000, time_factor=1.0
[0.618559, 0.999959, -0.785738, -0.00903877, 0.98768, 0.999999, 0.156486, 0.00157131, …]
shape [8, 4] · float32 · Tensor
2 implementations
2 papers
d1b6df964f7a
recorded values identical

pretrained_diffusion/unet.py a4255e26
recorded timesteps:1/float/float32; dim=8, max_period=10000, dtype=torch.float32

Protein/codes/nn/models.py df97c543
recorded timesteps:1/float/float32; dim=8, max_period=10000
[0.618559, 0.995918, 0.999959, 1, -0.785738, -0.0902659, -0.00903877, -0.000903889, …]
shape [8, 8] · float32 · Tensor
1 implementation
2 papers
a4027afbb776
one code sha held from 2 papers' repositories
library/flux_models.py 0a1aadb0
recorded t:1/float/float32; dim=4, max_period=10000, time_factor=1000.0
[0.62964, -0.926465, 0.776887, -0.376382, 0.998657, -0.000518366, 0.0518138, 1, …]
shape [8, 4] · float32 · Tensor

Bucket 2 of 2: arg 1: rank 1, kind int, dtype int64

9 implementations from 12 papers share this bucket (rank, kind, dtype of each array argument, positional; each member's recorded signature, argument name included, is shown under it); 3 distinct outputs, largest cluster first. Values are the first 8 of the recorded output, flattened.

Cluster (same digest to six decimals)MembersShared output on this bucket's input
7 implementations
11 papers
5bee047eaa31
recorded values identical
one code sha held from 3 papers' repositories
libs/uvit.py b494dd29
recorded timesteps:1/int/int64; dim=4, max_period=10000
one code sha held from 2 papers' repositories
src/models/polygon_models/polygon_net.py 12d88e58
recorded timesteps:1/int/int64; dim=4, max_period=10000
one code sha held from 2 papers' repositories
models/uvit.py 270cd726
recorded timesteps:1/int/int64; dim=4, max_period=10000

model.py 226b182e
recorded t:1/int/int64; dim=4, max_period=10000

ddpm/unet.py 8ceaa69a
recorded timesteps:1/int/int64; dim=4, max_period=10000

cldm/cldm.py 9df11570
recorded timesteps:1/int/int64; dim=4, max_period=10000, repeat_only=False

utils/models/mono.py dd00f566
recorded timesteps:1/int/int64; dim=4, max_period=10000
[-0.416147, 0.9998, 0.909297, 0.0199987, 0.540302, 0.99995, 0.841471, 0.00999983, …]
shape [8, 4] · float32 · Tensor
1 implementation
1 paper
3f9eb471628a

mr_traj/models.py 61ee83ed
recorded timesteps:1/int/int64; dim=4
[0.909297, 0.0002, -0.416147, 1, 0.841471, 0.0001, 0.540302, 1, …]
shape [8, 4] · float32 · Tensor
1 implementation
1 paper
e0a454ec42c0

model.py 88867f4d
recorded timesteps:1/int/int64; dim=8, max_period=10000
[-0.416147, 0.980067, 0.9998, 0.999998, 0.909297, 0.198669, 0.0199987, 0.002, …]
shape [8, 8] · float32 · Tensor

Identical values to six decimals (the recorded digest) on the shared input are agreement on those inputs, not equivalence; where a cluster's members carry recorded values, the largest difference among them is shown under the cluster. Paper titles are the archive's archive 2025-07-28 where the paper is in the archive and the graph's where it was added by Syntology; papers with no page here are shown by their recorded paper id only. A paper count above the implementation count means one implementation (one code sha) is held from several papers' repositories and counts once. Per-sample status, licence and fingerprint records for each paper are on its paper page. JSON twin: /census/timestep-embedding.json.

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