Home › Census › apply_rotary_pos_emb
apply_rotary_pos_emb
apply_rotary_pos_emb: 14 implementations from 15 papers ran on one shared input (census 2026-09-22, battery b986f7e04d79); they produced 8 distinct outputs across 7 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:
- 2 more implementations of this name were examined but did not run on the shared input (EinopsError 1, RuntimeError 1).
- No implementation of this name ran on its own fixture arguments only: every one that ran took an array argument.
- Every compared output was digested.
Bucket 1 of 7: arg 1: rank 3, kind float, dtype float32 · arg 2: rank 2, kind float, dtype float32
3 implementations from 3 papers share this bucket (rank, kind, dtype of each array argument, positional; each member's recorded signature, argument name included, is shown under it); 1 distinct output, largest cluster first. Values are the first 8 of the recorded output, flattened.
| Cluster (same digest to six decimals) | Members | Shared output on this bucket's input |
|---|---|---|
| 3 implementations 3 papers 316ba939c9ef recorded values identical |
block_recurrent_transformer_pytorch/block_recurrent_transformer_pytorch.py a1f5fb76 recorded t:3/float/float32, pos:2/float/float32; scale=1.0 retro_pytorch/retro_pytorch.py a6120b02 recorded t:3/float/float32, freqs:2/float/float32
models/framework.py f1c3b8a1 recorded t:3/float/float32, freqs:2/float/float32 |
[0.0582496, -1.37149, -0.962959, -0.441856, 0.794713, -2.18118, 0.506401, -0.460238, …] shape [2, 4, 8] · float32 · Tensor |
Bucket 2 of 7: arg 1: rank 3, kind float, dtype float32 · arg 2: rank 3, kind float, dtype float32
3 implementations from 3 papers share this bucket (rank, kind, dtype of each array argument, positional; each member's recorded signature, argument name included, is shown under it); 1 distinct output, largest cluster first. Values are the first 8 of the recorded output, flattened.
| Cluster (same digest to six decimals) | Members | Shared output on this bucket's input |
|---|---|---|
| 3 implementations 3 papers 5ee0be11597f recorded values identical |
perceiver_ar_pytorch/perceiver_ar_pytorch.py 746466b4 recorded pos:3/float/float32, t:3/float/float32 block_recurrent_transformer/transformer.py a0a60756 recorded t:3/float/float32, freqs:3/float/float32
libs/factorization_module.py c3acddf9 recorded t:3/float/float32, freqs:3/float/float32 |
[-0.442341, 1.23758, -0.710121, -0.346965, 0.0149312, -1.86188, 0.0993106, -0.257735, …] shape [2, 4, 8] · float32 · Tensor |
Bucket 3 of 7: arg 1: rank 2, kind float, dtype float32 · arg 2: rank 2, kind float, dtype float32
2 implementations from 1 paper share this bucket (rank, kind, dtype of each array argument, positional; each member's recorded signature, argument name included, is shown under it); 1 distinct output, largest cluster first. Values are the first 8 of the recorded output, flattened.
| Cluster (same digest to six decimals) | Members | Shared output on this bucket's input |
|---|---|---|
| 2 implementations 1 paper 775c5948d5b3 recorded values identical |
palm_pytorch/palm_pytorch.py 6d7a8eee recorded pos:2/float/float32, t:2/float/float32 coca_pytorch/coca_pytorch.py fd592af9 recorded pos:2/float/float32, t:2/float/float32 |
[0.464521, -1.08653, 0.726848, -0.167397, -1.97444, 0.785187, -0.815366, 0.106917, …] shape [4, 8] · float32 · Tensor |
Bucket 4 of 7: arg 1: rank 3, kind float, dtype float32 · arg 2: rank 3, kind float, dtype float32 · arg 3: rank 3, kind float, dtype float32
2 implementations from 2 papers share this bucket (rank, kind, dtype of each array argument, positional; each member's recorded signature, argument name included, is shown under it); 1 distinct output, largest cluster first. Values are the first 8 of the recorded output, flattened.
| Cluster (same digest to six decimals) | Members | Shared output on this bucket's input |
|---|---|---|
| 2 implementations 2 papers 30f0088025a6 recorded values identical |
src/omnisleep/models.py 0f92704b recorded x:3/float/float32, cos:3/float/float32, sin:3/float/float32 hacl/llava/model/language_model/llava.py 576e999c recorded x:3/float/float32, cos:3/float/float32, sin:3/float/float32; offset=0 |
[0.932535, 5.24648, 1.15641, -0.0317386, -0.112059, -0.355955, 0.543876, 0.243666, …] shape [2, 4, 8] · float32 · Tensor |
Bucket 5 of 7: arg 1: rank 4, kind float, dtype float32
2 implementations from 4 papers share this bucket (rank, kind, dtype of each array argument, positional; each member's recorded signature, argument name included, is shown under it); 2 distinct outputs, largest cluster first. Values are the first 8 of the recorded output, flattened.
| Cluster (same digest to six decimals) | Members | Shared output on this bucket's input |
|---|---|---|
| 1 implementation 3 papers b83d2fdbcca5 |
codegen1/jaxformer/hf/codegen/modeling_codegen.py 2f661886 recorded x:4/float/float32; sincos=(tensor([[ 1.1404, -0.0899], [ 0.7298, -1.8453], , offset=0 |
[-2.42357, -0.691662, -1.63137, -1.44428, -2.41247, 1.76109, 0.39959, -0.970987, …] shape [2, 3, 4, 4] · float32 · Tensor |
| 1 implementation 1 paper e7c45f49ac2b |
apicoder/CodeGenAPI/nl2code/modeling_codegen.py e206028c recorded x:4/float/float32; sincos=(tensor([[ 0.3704, 1.4565], [ 0.9398, 0.7748], , offset=0 |
[0.133132, 0.606544, -3.49805, 0.66437, 0.663895, 0.317557, -0.855447, -1.48741, …] shape [2, 3, 4, 4] · float32 · Tensor |
Bucket 6 of 7: arg 1: rank 2, kind float, dtype float32 · arg 2: rank 2, kind float, dtype float32 · arg 3: rank 2, kind float, dtype float32
1 implementation from 1 paper share this bucket (rank, kind, dtype of each array argument, positional; each member's recorded signature, argument name included, is shown under it); 1 distinct output, largest cluster first. Values are the first 8 of the recorded output, flattened.
| Cluster (same digest to six decimals) | Members | Shared output on this bucket's input |
|---|---|---|
| 1 implementation 1 paper 9606137f1b2a |
src/ctx_to_lora/modeling/text_to_lora.py 22f75dc2 recorded x:2/float/float32, cos:2/float/float32, sin:2/float/float32 |
[4.87697, 0.771957, 0.542136, 0.873127, -0.812365, 2.64238, 0.123743, -0.0675997, …] shape [4, 8] · float32 · Tensor |
Bucket 7 of 7: arg 1: rank 4, kind float, dtype float32 · arg 2: rank 4, kind float, dtype float32 · arg 3: rank 4, kind float, dtype float32
1 implementation from 2 papers share this bucket (rank, kind, dtype of each array argument, positional; each member's recorded signature, argument name included, is shown under it); 1 distinct output, largest cluster first. Values are the first 8 of the recorded output, flattened.
| Cluster (same digest to six decimals) | Members | Shared output on this bucket's input |
|---|---|---|
| 1 implementation 2 papers 094e8198ce32 |
src/models/controlnet.py 206c90f5 recorded x:4/float/float32, cos:4/float/float32, sin:4/float/float32 |
[-0.522791, -0.290921, 3.13684, 1.92422, 1.6022, 0.206306, -0.235056, 0.525854, …] shape [2, 3, 4, 4] · 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/apply-rotary-pos-emb.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