Papers › Memory Layers at Scale

Memory Layers at Scale

12 Dec 2024arXiv:2412.09764archive 2025-07-28

Vincent-Pierre Berges, Barlas Oğuz, Daniel Haziza, Wen-tau Yih, Luke Zettlemoyer, Gargi Ghosh

Memory layers use a trainable key-value lookup mechanism to add extra parameters to a model without increasing FLOPs. Conceptually, sparsely activated memory layers complement compute-heavy dense feed-forward layers, providing dedicated capacity to store and retrieve information cheaply. This work takes memory layers beyond proof-of-concept, proving their utility at contemporary scale. On downstream tasks, language models augmented with our improved memory layer outperform dense models with more than twice the computation budget, as well as mixture-of-expert models when matched for both compute and parameters. We find gains are especially pronounced for factual tasks. We provide a fully parallelizable memory layer implementation, demonstrating scaling laws with up to 128B memory parameters, pretrained to 1 trillion tokens, comparing to base models with up to 8B parameters.

PaperPDFCodeCode Syntology ran

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):

Code

Syntology 19 code samples harvested from 1 repository linked to this paper: 14 ran (of which 0 constructed an object rather than computing a result; 10 with no instrument failure: 1 honoured, 0 violated, 9 with no contract checked; 4 where Syntology's instrument failed) · 5 unverified. Where Syntology's instrument failed: 4 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 19 samples from 1 repository, 14 ran (10 with no instrument failure; 4 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.

Levels written between 2026-09-24 and 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

facebookresearch/memory officialmentioned in paperpytorchNOASSERTIONSyntology: 19 samples, 14 ran (4 instrument-failed) · 5 unverified report

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.

Code Syntology ran Syntology

19 samples harvested: 14 ran (of which 0 constructed an object rather than computing a result; 10 with no instrument failure: 1 honoured, 0 violated, 9 with no contract checked; 4 where Syntology's instrument failed) · 5 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · honoured contract
4ran · our draft was wrong
9ran
5unverified

Licence: 19 of the 19 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from facebookresearch/memory. “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.

facebookresearch/memory official repository 19 samples, 14 ran (4 instrument-failed) · 5 unverified
allclose validate_distributed_embeddingbag.py ran licence not identified · pointer only · 3eb31c04dd0c9897 · report fixture: signature a:Tensor/2/float, b:Tensor/2/float · returned Tensor · level recorded 2026-09-25
dataclass_from_dict lingua/args.py ran · our draft was wrong no licence file found · pointer only · e0ebc4027f1bd03b · report fixture: signature cls:callable/0/?, data:dict/0/?, strict:bool/0/bool · returned SimpleConfig · level recorded 2026-09-24
dataclass_to_dict lingua/args.py ran · our draft was wrong no licence file found · pointer only · 116b2f19f9a6148c · report fixture: signature dataclass_instance:SimpleConfig/0/? · returned dict · level recorded 2026-09-24
distribute_data_to_rank lingua/data.py ran licence not identified · pointer only · faca6fc3f7126a93 · report fixture: signature dataset_path:str/0/str, rank:int/0/int, world_size:int/0/int · returned dict · level recorded 2026-09-25
find_and_sanitize_chunks lingua/data.py ran licence not identified · pointer only · 3c8fff70712ff33a · report fixture: signature dataset_path:str/0/str, world_size:int/0/int · returned list · level recorded 2026-09-25
flatten_dict in 72 papers lingua/args.py ran · our draft was wrong no licence file found · pointer only · 15be2e56e073a228 · report fixture: signature d:dict/0/?, parent_key:str/0/str, sep:str/0/str · returned dict · level recorded 2026-09-24
get_empty_buffer_state lingua/data.py ran · our draft was wrong no licence file found · pointer only · bfdcca22e67a0a67 · report fixture: signature start_token:int/0/int, states:list/1/dict · returned dict · level recorded 2026-09-24
get_num_params in 15 papers lingua/metrics.py ran · honoured contract no licence file found · pointer only · 81d8582d1fab7bda · report fixture: signature model:callable/0/? · returned int · level recorded 2026-09-25
log_stats lingua/probe.py ran fingerprinted no licence file found · pointer only · 22fbaa2caf605450 · report fixture: signature x:Tensor/2/float, name:str/0/str · returned Tensor · level recorded 2026-09-25
lr_cosine lingua/optim.py ran licence not identified · pointer only · 758fe7262f8fa41e · report fixture: signature step:int/0/int, warmup:int/0/int, n_steps:int/0/int, cycle_length:float/0/float, theta:float/0/float, min_ratio:float/0/float · returned float · level recorded 2026-09-25
lr_inv_sqrt lingua/optim.py ran fingerprinted no licence file found · pointer only · 90403d20388a5636 · report fixture: signature step:int/0/int, warmup:int/0/int, exp_factor:float/0/float, min_ratio:float/0/float · returned float · level recorded 2026-09-25
lr_linear in 6 papers lingua/optim.py ran no licence file found · pointer only · 4e2454a79a8338d0 · report fixture: signature step:int/0/int, warmup:int/0/int, n_steps:int/0/int, min_ratio:float/0/float · returned float · level recorded 2026-09-25
named_replace lingua/float8.py ran no licence file found · pointer only · aaa4b37e564944fe · report fixture: signature fn:callable/0/?, module:callable/0/?, name:str/0/str · returned SimpleModel · level recorded 2026-09-25
scale in 38 papers lingua/float8.py ran no licence file found · pointer only · 2848341791f9a450 · report fixture: signature t:Tensor/2/float, amax_t:Tensor/1/float, dtype_t:dtype/0/? · returned tuple · level recorded 2026-09-25
consolidate_checkpoints in 6 papers lingua/checkpoint.py unverified licence not identified · pointer only · be7ab0634167f84c · report
dist_max in 7 papers lingua/distributed.py unverified no licence file found · pointer only · 9caf44ff8e00b1dc · report
dist_mean in 5 papers lingua/distributed.py unverified no licence file found · pointer only · d09c7072d82024f4 · report
get_device_mesh in 6 papers lingua/distributed.py unverified licence not identified · pointer only · eae71867f83f5695 · report
matmul in 24 papers lingua/float8.py unverified no licence file found · pointer only · 30e68c9cadfddd32 · report

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

BASE

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