SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2411.12892 · 2024

Selective Attention: Enhancing Transformer through Principled Context Control

arXiv · PDF · Open in the Atlas

Code that ran

3 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 2 violated it and 0 have no contract verdict; 0 failures of our instrument; 1 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2411.12892.json · badge: /api/ran/2411.12892.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
umich-sota/selective_attention canonical (classifier label)
classifier basis: classifier_2025
3 ran · 1 no recorded run
FunctionStatusWhere it lives
apply_rope Ran
ran, violated the contract we drafted
umich-sota/selective_attention/lit_gpt/model.py
code served (permissive licence) · get_code("aa32c6faaa562b69")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 3, dtype family float, shape (2, 3, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Construct tensors for apply_rope function
# x: input tensor with shape (batch, seq_len, head_size)
# The function expects x to have a last dimension (head_size) that can be split in half
batch_size = 2
seq_len = 3
head_size = 4

x = torch.randn(batch_size, seq_len, head_size)

# cos and sin tensors should have compatible shape for broadcasting
# They should be able to broadcast with x
cos = torch.randn(batch_size, seq_len, head_size)
sin = torch.randn(batch_size, seq_len, head_size)

ARGS = {
    "x": x,
    "cos": cos,
    "sin": sin,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 3, 4) — shape (2, 3, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
build_mask_cache Ran
ran, violated the contract we drafted
umich-sota/selective_attention/lit_gpt/model.py
code served (permissive licence) · get_code("7ee1c3ba71156110")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 4, dtype family float, shape (1, 1, 8, 8) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

ARGS = {
    "max_seq_length": 8,
    "device": None,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 4, dtype family float, shape (1, 1, 8, 8) — shape (1, 1, 8, 8) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_longest_seq_length Ran
ran, honoured the contract we drafted
umich-sota/selective_attention/finetune/full.py
code served (permissive licence) · get_code("7f76ef00622697f4")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

# Construct minimal valid arguments for get_longest_seq_length
# The function expects a list of dicts, where each dict has 'input_ids' key
# with a value that supports len()

torch.manual_seed(0)

# Create sample data with different sequence lengths
data = [
    {'input_ids': [1, 2, 3]},
    {'input_ids': [4, 5, 6, 7, 8]},  # longest sequence
    {'input_ids': [9, 10]},
]

ARGS = {
    'data': data
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family int — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
build_rope_cache Not yet run
unverified (harvested, no recorded run)
umich-sota/selective_attention/lit_gpt/model.py
code served (permissive licence) · get_code("c0cabd60c81402a5")
How it ran

no invocation recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 0 later) and 1 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
build_rope_cache 6 other papers
6 dated earlier
2405.20340 (2024-05)
MotionLLM: Understanding Human Behaviors from Human Motions and Videos
2406.06046 (2024-06)
MATES : Model-Aware Data Selection for Efficient Pretraining with Data Influence Models
2406.10209 (2024-06)
Be like a Goldfish, Don't Memorize! Mitigating Memorization in Generative LLMs
and 3 more not named here
apply_rope 5 other papers
5 dated earlier
2406.06046 (2024-06)
MATES : Model-Aware Data Selection for Efficient Pretraining with Data Influence Models
2406.10209 (2024-06)
Be like a Goldfish, Don't Memorize! Mitigating Memorization in Generative LLMs
2408.16725 (2024-08)
Mini-Omni: Language Models Can Hear, Talk While Thinking in Streaming
and 2 more not named here
build_mask_cache 5 other papers
5 dated earlier
2406.06046 (2024-06)
MATES : Model-Aware Data Selection for Efficient Pretraining with Data Influence Models
2406.10209 (2024-06)
Be like a Goldfish, Don't Memorize! Mitigating Memorization in Generative LLMs
2408.16725 (2024-08)
Mini-Omni: Language Models Can Hear, Talk While Thinking in Streaming
and 2 more not named here

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

The attention mechanism within the transformer architecture enables the model to weigh and combine tokens based on their relevance to the query. While self-attention has enjoyed major success, it notably treats all queries $q$ in the same way by applying the mapping $V^\top\text{softmax}(Kq)$, where $V,K$ are the value and key embeddings respectively. In this work, we argue that this uniform treatment hinders the ability to control contextual sparsity and relevance. As a solution, we introduce the $\textit{Selective Self-Attention}$ (SSA) layer that augments the softmax nonlinearity with a principled temperature scaling strategy. By controlling temperature, SSA adapts the contextual sparsity of the attention map to the query embedding and its position in the context window. Through theory and experiments, we demonstrate that this alleviates attention dilution, aids the optimization process, and enhances the model's ability to control softmax spikiness of individual queries. We also incorporate temperature scaling for value embeddings and show that it boosts the model's ability to suppress irrelevant/noisy tokens. Notably, SSA is a lightweight method which introduces less than 0.5% new parameters through a weight-sharing strategy and can be fine-tuned on existing LLMs. Extensive empirical evaluations demonstrate that SSA-equipped models achieve a noticeable and consistent accuracy improvement on language modeling benchmarks.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2411.12892")
get_code_for_paper("2411.12892")
have("2411.12892")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2411.12892.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2411.12892.svg)](https://syntology.ai/paper/2411.12892)

Connect an agent — have() is free.