Papers › A is for Absorption: Studying Feature Splitting and Absorption in Sparse Autoencoders

A is for Absorption: Studying Feature Splitting and Absorption in Sparse Autoencoders

22 Sep 2024arXiv:2409.14507archive 2025-07-28

David Chanin, James Wilken-Smith, Tomáš Dulka, Hardik Bhatnagar, Joseph Bloom

Sparse Autoencoders (SAEs) have emerged as a promising approach to decompose the activations of Large Language Models (LLMs) into human-interpretable latents. In this paper, we pose two questions. First, to what extent do SAEs extract monosemantic and interpretable latents? Second, to what extent does varying the sparsity or the size of the SAE affect monosemanticity / interpretability? By investigating these questions in the context of a simple first-letter identification task where we have complete access to ground truth labels for all tokens in the vocabulary, we are able to provide more detail than prior investigations. Critically, we identify a problematic form of feature-splitting we call feature absorption where seemingly monosemantic latents fail to fire in cases where they clearly should. Our investigation suggests that varying SAE size or sparsity is insufficient to solve this issue, and that there are deeper conceptual issues in need of resolution.

PaperPDFCodeCode Syntology ranMethod as code

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.

Syntology holds a reference implementation of a metric to detect absorption in SAEs, generated from the paper text (not the authors' code), at level 3 on Syntology's ladder: V2 “two implementations GENERATED from the paper's signature and spec agree numerically (1e-6) on fixture inputs” “neither of the two ever read the authors' own code, so V2 measures how unambiguously the paper specifies the method, not fidelity to the authors' implementation (measured 2026-09-11, commit 75eaa77: cross_checked_by is a second LLM on 2,628 of 2,831 served samples)”; V3 “paper-derived property tests pass on both of those same two generations.” (quoted from Syntology's server card). No rung compares this code against the authors' implementation. For agents: get_reference_implementation(name='a metric to detect absorption in SAEs') (how to connect); the free check have('2409.14507') needs no token.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 10 code samples harvested from 1 repository linked to this paper: 9 ran (of which 0 constructed an object rather than computing a result; 6 with no instrument failure: 1 honoured, 0 violated, 5 with no contract checked; 3 where Syntology's instrument failed) · 1 unverified. Where Syntology's instrument failed: 3 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 10 samples from 1 repository, 9 ran (6 with no instrument failure; 3 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 on 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

lasr-spelling/sae-spelling officialmentioned in papermentioned on GitHubpytorchMITSyntology: 10 samples, 9 ran (3 instrument-failed) · 1 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

10 samples harvested: 9 ran (of which 0 constructed an object rather than computing a result; 6 with no instrument failure: 1 honoured, 0 violated, 5 with no contract checked; 3 where Syntology's instrument failed) · 1 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
3ran · our draft was wrong
5ran
1unverified

Licence: 0 of the 10 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 lasr-spelling/sae-spelling. “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.

lasr-spelling/sae-spelling official repository 10 samples, 9 ran (3 instrument-failed) · 1 unverified
dtype_to_str sae_spelling/experiments/common.py ran · our draft was wrong MIT (permissive) · dcaa80bbfa602dce · report fixture: signature dtype:dtype/0/? · returned str · level recorded 2026-09-24
extract_grad sae_spelling/torch_utils.py ran · our draft was wrong MIT (permissive) · 7fb9a9df9e2f3f9a · report fixture: signature tensor:Tensor/2/float · returned Tensor · level recorded 2026-09-24
first_letter sae_spelling/prompting.py ran MIT (permissive) · 7fb0d3f1a609202f · report fixture: signature word:str/0/str, prefix:str/0/str, capitalize:bool/0/bool, ignore_leading_space:bool/0/bool, ignore_non_alpha_chars:bool/0/bool · returned str · level recorded 2026-09-24
flip_dict sae_spelling/util.py ran · our draft was wrong MIT (permissive) · 102789285bbb6108 · report fixture: signature d:dict/0/? · returned dict · level recorded 2026-09-24
get_alpha_tokens sae_spelling/vocab.py ran MIT (permissive) · c4c34bce5b14a047 · report fixture: signature tokenizer:callable/0/?, allow_leading_space:bool/0/bool, replace_special_chars:bool/0/bool · returned list · level recorded 2026-09-24
get_tokens in 40 papers sae_spelling/vocab.py ran MIT (permissive) · 927eb86571fdf5f4 · report fixture: signature tokenizer:callable/0/?, filter:callable/0/?, replace_special_chars:bool/0/bool · returned list · level recorded 2026-09-24
last_letter sae_spelling/prompting.py ran MIT (permissive) · 5da8ce88eae79932 · report fixture: signature word:str/0/str, prefix:str/0/str, capitalize:bool/0/bool, ignore_leading_space:bool/0/bool, ignore_non_alpha_chars:bool/0/bool · returned str · level recorded 2026-09-24
listify in 14 papers sae_spelling/util.py ran · honoured contract fingerprinted MIT (permissive) · 4ea90f6ac63c39b8 · report fixture: signature item:int/0/int · returned list · level recorded 2026-09-24
spelling sae_spelling/prompting.py ran MIT (permissive) · 4b3741b6550dd634 · report fixture: signature word:str/0/str, separator:str/0/str, prefix:str/0/str, capitalize:bool/0/bool, ignore_leading_space:bool/0/bool, ignore_non_alpha_chars:bool/0/bool · returned str · level recorded 2026-09-24
get_common_word_tokens sae_spelling/vocab.py unverified MIT (permissive) · e61d11c374aa3983 · report

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. For a generated assembly, an independent peer is a second implementation generated separately from the same paper text; neither implementation read the authors' code. At level 2 the two agreed numerically on the fixture's inputs; level 3 is recorded when property tests derived from the paper also passed on both, and each level 3 card prints how many its report lists. A generated assembly's label was set when its code was generated for the method, and no model judged it; its card prints it as "by construction" or "not judged by a model". Where an assembly's phrase has no constructed figure, a brick that ran is not among the harvested samples this build reads, and whether a brick's run only constructed an object is read from that brick's harvested sample, so the figure was not counted; it does not say there were none.

1 assembly on this paper: 0 harvested, 1 generated.

generated Assembly for a metric to detect absorption in SAEs (graph method: a metric to detect absorption in SAEs), generated from the paper text, not the repository's code. 1 brick; individually on a fixture: 1 ran (1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified. Ran end-to-end against an independent peer (level 3; 5 property tests passed on both).

label matches by construction: the code was generated for this method (not judged by a model)

  1. compute_ablation_effect entry · level 3 · ran · honoured contract · afc3999247af12abcdb66f09639213baa85279dab1f86842870a7141d51633ec

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2409.14507") returns these assemblies with each brick's role and wiring (how to connect).

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

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