Papers › RAVEL: Evaluating Interpretability Methods on Disentangling Language Model Representations

RAVEL: Evaluating Interpretability Methods on Disentangling Language Model Representations

27 Feb 2024arXiv:2402.17700archive 2025-07-28

Jing Huang, Zhengxuan Wu, Christopher Potts, Mor Geva, Atticus Geiger

Individual neurons participate in the representation of multiple high-level concepts. To what extent can different interpretability methods successfully disentangle these roles? To help address this question, we introduce RAVEL (Resolving Attribute-Value Entanglements in Language Models), a dataset that enables tightly controlled, quantitative comparisons between a variety of existing interpretability methods. We use the resulting conceptual framework to define the new method of Multi-task Distributed Alignment Search (MDAS), which allows us to find distributed representations satisfying multiple causal criteria. With Llama2-7B as the target language model, MDAS achieves state-of-the-art results on RAVEL, demonstrating the importance of going beyond neuron-level analyses to identify features distributed across activations. We release our benchmark at https://github.com/explanare/ravel.

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 17 code samples harvested from 1 repository linked to this paper: 12 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 0 honoured, 0 violated, 9 with no contract checked; 3 where Syntology's instrument failed) · 5 unverified. Where Syntology's instrument failed: 2 where our contract draft was wrong; 1 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: official repository: 17 samples from 1 repository, 12 ran (9 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.

explanare/ravel officialmentioned in paperpytorchMITSyntology: 17 samples, 12 ran (3 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

17 samples harvested: 12 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 0 honoured, 0 violated, 9 with no contract checked; 3 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.

2ran · our draft was wrong
1ran; the contract check could not run
9ran
5unverified

Licence: 1 of the 17 samples is pointer only, meaning Syntology does not serve that copy's text. For that sample, this paper's copy is under a permissive licence and cleared, but the stored copy of the same code that Syntology serves is not; the licence label's tooltip gives Syntology's recorded reason. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from explanare/ravel. “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.

explanare/ravel official repository 17 samples, 12 ran (3 instrument-failed) · 5 unverified
compute_cross_entropy_loss src/utils/metric_utils.py ran MIT (permissive) · 39d29ba2dce082a9 · report fixture: signature logits:Tensor/3/float, labels:Tensor/2/int, pad_token_id:int/0/int, next_n_tokens:int/0/int · returned Tensor · level recorded 2026-09-24
compute_disentangle_score src/utils/metric_utils.py ran MIT (permissive) · 618cb3bd9cfb4a14 · report fixture: signature log_data:dict/0/?, attribute_to_iso_tasks:dict/0/?, attribute_to_cause_tasks:dict/0/? · returned dict · level recorded 2026-09-24
compute_metrics in 208 papers src/utils/metric_utils.py ran MIT (permissive) · 5500053238475cfe · report fixture: signature eval_preds:list/1/Tensor, eval_labels:list/1/Tensor, pad_token_id:int/0/int, last_n_tokens:int/0/int · returned dict · level recorded 2026-09-24
compute_principal_component src/methods/pca.py ran; the contract check could not run (RAISES) MIT (permissive) · ebcb0699723a970c · report fixture: signature features:ndarray/2/float, n_components:int/0/int · returned dict · level recorded 2026-09-24
generate_batched src/utils/generation_utils.py ran MIT (permissive) · e7618272c2024815 · report fixture: signature pretrained_model:callable/0/?, tokenizer:callable/0/?, all_prompts:list/1/str, max_length:int/0/int, prompt_max_length:int/0/int, max_new_tokens:int/0/int, sample_n:int/0/int, batch_size:int/0/int · returned list · level recorded 2026-09-24
get_score in 56 papers src/methods/linear_adversarial_probe.py ran MIT (permissive) · 4637704289ff6a8a · report fixture: signature X_train:ndarray/2/float, y_train:ndarray/1/int, X_dev:ndarray/2/float, y_dev:ndarray/1/int, P:ndarray/2/float, rank:int/0/int · returned tuple · level recorded 2026-09-24
init_sgd_classifier src/methods/linear_adversarial_probe.py ran · our draft was wrong MIT (permissive) · 9a0c2776f5ac2bc0 · report fixture: signature params:dict/0/? · returned SGDClassifier · level recorded 2026-09-24
kept_first_n_label_token src/utils/dataset_utils.py ran MIT (permissive) · f648fddc170ec6d8 · report fixture: signature x:dict/0/?, first_n:int/0/int, padding_offset:int/0/int · returned dict · level recorded 2026-09-24
kept_first_n_label_token_multitask src/utils/dataset_utils.py ran MIT (permissive) · f404f9aad2c1b27e · report fixture: signature x:dict/0/?, cause_tasks:list/1/str, first_n:int/0/int, padding_offset:int/0/int · returned dict · level recorded 2026-09-24
preproc_tokenize src/utils/dataset_utils.py ran MIT (permissive) · 39872fcdf40c8524 · report fixture: signature tokenizer:callable/0/?, max_input_length:int/0/int, max_output_length:int/0/int, examples:dict/0/?, input_feature:str/0/str, label_feature:str/0/str, extra_input_to_tokenize:list/1/str, extra_label_to_tokenize:list/1/str · returned dict · level recorded 2026-09-24
remove_invalid_token_id src/utils/intervention_utils.py ran · our draft was wrong fingerprinted MIT (permissive) · fe8f9421a66f408f · report fixture: signature token_ids:Tensor/2/int, pad_id:int/0/int · returned Tensor · level recorded 2026-09-24
select_features_with_classifier src/methods/select_features.py ran MIT (permissive) · be76384e0b33672a · report fixture: signature featurizer:callable/0/?, inputs:Tensor/2/float, labels:ndarray/1/int, coeff:NoneType/0/? · returned dict · level recorded 2026-09-24
gen_context_test_split src/utils/generate_ravel_instance.py unverified MIT (permissive) · 8240c646692f8698 · report
gen_entity_test_split src/utils/generate_ravel_instance.py unverified MIT (permissive) · 58779741197dded2 · report
gen_train_split src/utils/generate_ravel_instance.py unverified MIT (permissive) · 8289a1e4195d682d · report
symmetric in 6 papers src/methods/linear_adversarial_probe.py unverified MIT (permissive) here; the stored copy is pointer only · 5eb6f7201a4d2f1c · report
train_intervention_step src/utils/intervention_utils.py unverified MIT (permissive) · 01750c9c0ff161d4 · report

Tasks

AttributeLanguage ModelingLanguage Modelling

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