Papers › Counterfactual Token Generation in Large Language Models

Counterfactual Token Generation in Large Language Models

25 Sep 2024arXiv:2409.17027archive 2025-07-28

Ivi Chatzi, Nina Corvelo Benz, Eleni Straitouri, Stratis Tsirtsis, Manuel Gomez-Rodriguez

"Sure, I am happy to generate a story for you: Captain Lyra stood at the helm of her trusty ship, the Maelstrom's Fury, gazing out at the endless sea. [...] Lyra's eyes welled up with tears as she realized the bitter truth - she had sacrificed everything for fleeting riches, and lost the love of her crew, her family, and herself." Although this story, generated by a large language model, is captivating, one may wonder -- how would the story have unfolded if the model had chosen "Captain Maeve" as the protagonist instead? We cannot know. State-of-the-art large language models are stateless -- they maintain no internal memory or state. Given a prompt, they generate a sequence of tokens as an output using an autoregressive process. As a consequence, they cannot reason about counterfactual alternatives to tokens they have generated in the past. In this work, our goal is to enhance them with this functionality. To this end, we develop a causal model of token generation that builds upon the Gumbel-Max structural causal model. Our model allows any large language model to perform counterfactual token generation at almost no cost in comparison with vanilla token generation, it is embarrassingly simple to implement, and it does not require any fine-tuning nor prompt engineering. We implement our model on Llama 3 8B-Instruct and Ministral-8B-Instruct and conduct a qualitative and a quantitative analysis of counterfactually generated text. We conclude with a demonstrative application of counterfactual token generation for bias detection, unveiling interesting insights about the model of the world constructed by large language models.

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 10 code samples harvested from 1 repository linked to this paper: 10 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 1 honoured, 0 violated, 8 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified. Where Syntology's instrument failed: 1 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: official repository: 10 samples from 1 repository, 10 ran (9 with no instrument failure; 1 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. 1 of the 10 samples that ran has no recorded date.

networks-learning/counterfactual-llms officialmentioned in papermentioned on GitHubpytorchSyntology: 10 samples, 10 ran (1 instrument-failed) · 0 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: 10 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 1 honoured, 0 violated, 8 with no contract checked; 1 where Syntology's instrument failed) · 0 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
1ran; the contract check could not run
8ran

Licence: 10 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 networks-learning/counterfactual-llms. “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. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

networks-learning/counterfactual-llms official repository 10 samples, 10 ran (1 instrument-failed) · 0 unverified
apply_rotary_emb in 121 papers src/mistral-inference/moe_one_file_ref.py ran no licence file found · pointer only · 22426d7eecb70410 · report fixture: signature xq:Tensor/3/float, xk:Tensor/3/float, freqs_cis:Tensor/2/complex · returned tuple · level recorded 2026-09-24
fix_income src/bias.py ran · honoured contract fingerprinted no licence file found · pointer only · 7e3a2dbb3c9579a1 · report fixture: signature value:str/0/str · returned int · level recorded 2026-09-24
get_fig_dim src/utils.py ran fingerprinted no licence file found · pointer only · 1c1a16c4896a4f25 · report fixture: signature width:float/0/float, fraction:float/0/float, aspect_ratio:float/0/float · returned tuple · level recorded 2026-09-24
get_where_to_intervene_dict src/bias.py ran no licence file found · pointer only · 7c2a58be8462ca91 · report fixture: signature token_list:dict/0/?, intervention_keys:set/0/?, model_family:str/0/str · returned dict · level recorded 2026-09-24
get_where_to_intervene_dict_direct src/bias.py ran no licence file found · pointer only · 58d741b58f7d0059 · report fixture: signature token_list:dict/0/?, intervention_keys:list/1/str, model_family:str/0/str · returned dict · level recorded 2026-09-24
precompute_freqs_cis in 112 papers src/mistral-inference/moe_one_file_ref.py ran no licence file found · pointer only · 99e0ca21ba709107 · report fixture: signature dim:int/0/int, end:int/0/int, theta:float/0/float · returned Tensor · level recorded 2026-09-24
repeat_kv in 244 papers
census: 15 implementations from 60 papers, 2 distinct outputs on one shared input
src/mistral-inference/moe_one_file_ref.py ran no licence file found · pointer only · 3fdd4f66367ca608 · report fixture: signature keys:Tensor/4/float, values:Tensor/4/float, repeats:int/0/int · returned tuple · level recorded 2026-09-24
repeat_kv in 244 papers
census: 15 implementations from 60 papers, 2 distinct outputs on one shared input
src/mistral-inference/one_file_ref.py ran no licence file found · pointer only · e8e39c9933cb102c · report fixture: signature keys:Tensor/3/float, values:Tensor/3/float, repeats:int/0/int · returned tuple · level recorded 2026-09-24
reshape_for_broadcast in 90 papers
census: 6 implementations from 10 papers, 1 distinct output on one shared input
src/llama3/llama/model.py ran; the contract check could not run (RAISES) · metamorphic tier: invariant fingerprinted no licence file found · pointer only · 70bf6ebaafd266c4 · report fixture: signature freqs_cis:Tensor/2/float, x:Tensor/4/float · returned Tensor · level recorded, date not recorded
sample_top_p in 35 papers src/llama3/llama/generation.py ran no licence file found · pointer only · 841e0eb44df294a1 · report fixture: signature probs:Tensor/2/float, p:float/0/float, rng:Generator/0/? · returned Tensor · level recorded 2026-09-24

Tasks

Bias DetectionLanguage ModelingLanguage ModellingLarge Language ModelPrompt Engineering

1 archive task tag without a task page not shown.

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

AMLLaMA

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