SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2305.13245 · EMNLP · 2023

GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints

Joshua Ainslie, Yury Zemlyanskiy, Sumit Sanghai, Michiel De Jong, James Lee-Thorp, Federico Lebrón

arXiv · PDF · Open in the Atlas

Code that ran

4 of 5 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (2 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 1 failure 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/2305.13245.json · badge: /api/ran/2305.13245.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
knotgrass/attention community
Papers with Code archive
3 ran · 0 no recorded run
fkodom/grouped-query-attention-pytorch community
Papers with Code archive
1 ran · 1 no recorded run
FunctionStatusWhere it lives
Attention Ran
ran on a synthesized input · constructed an object; no result was computed
knotgrass/attention/attn/attention.py
pointer only (licence: NONE) · get_code("7bd4605898291f6a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, knotgrass/attention when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

GroupedQueryAttention Ran
ran on a synthesized input
knotgrass/attention/attn/attention.py
pointer only (licence: NONE) · get_code("959111f20c6f8476")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, knotgrass/attention when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

MultiQueryAttention Ran
ran on a synthesized input · constructed an object; no result was computed
knotgrass/attention/attn/attention.py
pointer only (licence: NONE) · get_code("edf29ee77f89d029")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, knotgrass/attention when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

scaled_dot_product_gqa Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
fkodom/grouped-query-attention-pytorch/grouped_query_attention_pytorch/attention.py
code served (permissive licence) · get_code("aa68f534d467d1bd")
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) and returned tuple. The image it ran in was not recorded. If your copy returns the same type 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.functional as F
from einops import einsum, rearrange

torch.manual_seed(0)

# Tiny tensor dimensions: batch=2, seq_len_query=3, seq_len_key=4, heads_query=4, heads_key=2, dim=8
batch_size = 2
seq_len_query = 3
seq_len_key = 4
num_heads_query = 4
num_heads_key = 2
embedding_dim = 8

# Create query, key, value tensors with shape (b, n/s, h, d)
query = torch.randn(batch_size, seq_len_query, num_heads_query, embedding_dim)
key = torch.randn(batch_size, seq_len_key, num_heads_key, embedding_dim)
value = torch.randn(batch_size, seq_len_key, num_heads_key, embedding_dim)

# Create a 3D mask (b, n, s)
mask_3d = torch.ones(batch_size, seq_len_query, seq_len_key, dtype=torch.bool)

# Create a 2D mask (b, s)
mask_2d = torch.ones(batch_size, seq_len_key, dtype=torch.bool)

ARGS = {
    "query": query,
    "key": key,
    "value": value,
    "dropout": 0.1,
    "scale": None,
    "mask": mask_3d,
    "is_causal": None,
    "need_weights": True,
    "average_attn_weights": False,
    "force_grouped": False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
einops, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
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
MultiheadGQA Not yet run
unverified (harvested, no recorded run)
fkodom/grouped-query-attention-pytorch/grouped_query_attention_pytorch/attention.py
code served (permissive licence) · get_code("5cb9f0bfab83e93d")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 4 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
Attention
brick of this paper's assembly: helper
1 other paper
1 dated earlier · 1 only as an assembly brick
1706.03762 (2017-06)
Attention Is All You Need

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

Multi-query attention (MQA), which only uses a single key-value head, drastically speeds up decoder inference. However, MQA can lead to quality degradation, and moreover it may not be desirable to train a separate model just for faster inference. We (1) propose a recipe for uptraining existing multi-head language model checkpoints into models with MQA using 5% of original pre-training compute, and (2) introduce grouped-query attention (GQA), a generalization of multi-query attention which uses an intermediate (more than one, less than number of query heads) number of key-value heads. We show that uptrained GQA achieves quality close to multi-head attention with comparable speed to MQA.

For agents

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

get_harvested_code_for_paper("2305.13245")
get_code_for_paper("2305.13245")
have("2305.13245")

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

curl https://syntology.ai/api/ran/2305.13245.json

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

[![Syntology run record](https://syntology.ai/api/ran/2305.13245.svg)](https://syntology.ai/paper/2305.13245)

Connect an agent — have() is free.