Papers › Transformers are Sample-Efficient World Models

Transformers are Sample-Efficient World Models

1 Sep 2022arXiv:2209.00588archive 2025-07-28

Vincent Micheli, Eloi Alonso, François Fleuret

Deep reinforcement learning agents are notoriously sample inefficient, which considerably limits their application to real-world problems. Recently, many model-based methods have been designed to address this issue, with learning in the imagination of a world model being one of the most prominent approaches. However, while virtually unlimited interaction with a simulated environment sounds appealing, the world model has to be accurate over extended periods of time. Motivated by the success of Transformers in sequence modeling tasks, we introduce IRIS, a data-efficient agent that learns in a world model composed of a discrete autoencoder and an autoregressive Transformer. With the equivalent of only two hours of gameplay in the Atari 100k benchmark, IRIS achieves a mean human normalized score of 1.046, and outperforms humans on 10 out of 26 games, setting a new state of the art for methods without lookahead search. To foster future research on Transformers and world models for sample-efficient reinforcement learning, we release our code and models at https://github.com/eloialonso/iris.

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.

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

Code

Syntology 26 code samples harvested from 1 repository linked to this paper: 17 ran (of which 13 constructed an object rather than computing a result; 16 with no instrument failure: 1 honoured, 0 violated, 15 with no contract checked; 1 where Syntology's instrument failed) · 9 unverified. Where Syntology's instrument failed: 1 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 26 samples from 1 repository, 17 ran (16 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.

No sample is linked to this paper from 1 of its repositories that Syntology harvested for other papers:

Levels written between 2026-09-22 and 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

eloialonso/iris officialmentioned in papermentioned on GitHubpytorchSyntology: 26 samples, 17 ran (1 instrument-failed) · 9 unverified report
vmicheli/delta-iris mentioned on GitHubpytorchGPL-3.0Syntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Efficient World Models with Context-Aware Tokenization (arXiv:2406.19320; that paper's own run record: 6 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 1 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 1 unverified); nothing checks that they implement this paper's method. 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.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

26 samples harvested: 17 ran (of which 13 constructed an object rather than computing a result; 16 with no instrument failure: 1 honoured, 0 violated, 15 with no contract checked; 1 where Syntology's instrument failed) · 9 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · honoured contract
1ran · our draft was wrong
15ran
9unverified

Licence: 26 of the 26 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 eloialonso/iris. “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”. For 1 sample that ran, the graph records a signature this page does not show: for a class the fixture built without calling, the harness called the object with inputs it made up itself, so the recorded signature is the harness's probe, not the fixture's arguments. “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.

eloialonso/iris official repository 26 samples, 17 ran (1 instrument-failed) · 9 unverified
AssignWithoutInplaceCheck src/models/world_model.py ran GPL-3.0 (copyleft) · pointer only · 7f939e4da5014501 · report fixture: returned AssignWithoutInplaceCheck · level recorded 2026-09-22
AttnBlock in 22 papers src/models/world_model.py ran GPL-3.0 (copyleft) · pointer only · 9d1f4ad302b969e4 · report fixture: signature x:Tensor/4/float · returned Tensor · level recorded 2026-09-24
Cache in 5 papers src/models/world_model.py ran GPL-3.0 (copyleft) · pointer only · 256e6e8e5c0039da · report fixture: returned Cache · level recorded 2026-09-22
Downsample in 58 papers src/models/world_model.py ran · metamorphic tier: deterministic fingerprinted GPL-3.0 (copyleft) · pointer only · e8df5d8fcd9e9339 · report fixture: returned Downsample · level recorded 2026-09-22
Embedder in 18 papers src/models/world_model.py ran · metamorphic tier: deterministic GPL-3.0 (copyleft) · pointer only · 2def3130991aaf3c · report fixture: returned Embedder · level recorded 2026-09-22
EncoderDecoderConfig src/models/world_model.py ran · metamorphic tier: deterministic GPL-3.0 (copyleft) · pointer only · cd65aa1097b08332 · report fixture: returned EncoderDecoderConfig · level recorded 2026-09-22
Head in 14 papers src/models/world_model.py ran · metamorphic tier: deterministic GPL-3.0 (copyleft) · pointer only · b4868fbed7dce84a · report fixture: returned Head · level recorded 2026-09-22
KVCache in 5 papers src/models/world_model.py ran GPL-3.0 (copyleft) · pointer only · 93bd9c4b1f02db59 · report fixture: returned KVCache · level recorded 2026-09-22
KeysValues src/models/world_model.py ran GPL-3.0 (copyleft) · pointer only · fba3ef4043569d6f · report fixture: returned KeysValues · level recorded 2026-09-22
LossWithIntermediateLosses src/models/world_model.py ran GPL-3.0 (copyleft) · pointer only · d7a1b7c2fadc881a · report fixture: returned LossWithIntermediateLosses · level recorded 2026-09-22
Normalize in 154 papers src/models/world_model.py ran · our draft was wrong GPL-3.0 (copyleft) · pointer only · 41ffde6c773af0d6 · report fixture: signature in_channels:int/0/int · returned GroupNorm · level recorded 2026-09-22
Slicer src/models/world_model.py ran · metamorphic tier: deterministic GPL-3.0 (copyleft) · pointer only · 3e962c6d7a040002 · report fixture: returned Slicer · level recorded 2026-09-22
TokenizerEncoderOutput src/models/world_model.py ran · metamorphic tier: deterministic GPL-3.0 (copyleft) · pointer only · bf4dc4c0382a6bd0 · report fixture: returned TokenizerEncoderOutput · level recorded 2026-09-22
TransformerConfig src/models/world_model.py ran · metamorphic tier: deterministic GPL-3.0 (copyleft) · pointer only · 06c7be0a3042efdc · report fixture: returned TransformerConfig · level recorded 2026-09-22
Upsample in 77 papers src/models/world_model.py ran fingerprinted GPL-3.0 (copyleft) · pointer only · 53918c7471cbccdc · report fixture: signature x:Tensor/4/float · returned Tensor · level recorded 2026-09-24
WorldModelOutput src/models/world_model.py ran · metamorphic tier: deterministic GPL-3.0 (copyleft) · pointer only · 419935433daf0f1b · report fixture: returned WorldModelOutput · level recorded 2026-09-22
nonlinearity in 140 papers
census: 4 implementations from 11 papers, 1 distinct output on one shared input this sample: bucket 1 of 1, arg 1: rank 2, kind float, dtype float32
src/models/world_model.py ran · honoured contract fingerprinted GPL-3.0 (copyleft) · pointer only · 3f6efbafceb71819 · report fixture: signature x:Tensor/2/float · returned Tensor · level recorded 2026-09-22
Decoder in 210 papers src/models/world_model.py unverified GPL-3.0 (copyleft) · pointer only · 4e93d0d0e4839243 · report
Encoder in 278 papers src/models/world_model.py unverified GPL-3.0 (copyleft) · pointer only · aee0695eaeb766c4 · report
ResnetBlock in 49 papers src/models/world_model.py unverified GPL-3.0 (copyleft) · pointer only · caffc15eb259ebb0 · report
SelfAttention in 66 papers src/models/world_model.py unverified GPL-3.0 (copyleft) · pointer only · f16fd90513419812 · report
Tokenizer in 16 papers src/models/world_model.py unverified GPL-3.0 (copyleft) · pointer only · 79b0b0b918ab2ac4 · report
Transformer in 173 papers src/models/world_model.py unverified GPL-3.0 (copyleft) · pointer only · d65bf756384c37cd · report
WorldModel in 13 papers src/models/world_model.py unverified GPL-3.0 (copyleft) · pointer only · cfb3154cdccd72da · report
init_weights in 109 papers src/models/world_model.py unverified GPL-3.0 (copyleft) · pointer only · 7fc674de9a0a0a02 · 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. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it.

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

harvested Assembly for IRIS (graph method: IRIS). 26 bricks; individually on a fixture: 17 ran (of which 13 constructed an object rather than computing a result; 16 with no instrument failure: 1 honoured, 0 violated, 15 with no contract checked; 1 where Syntology's instrument failed) · 9 unverified. The assembly has not been run end-to-end.

  1. WorldModel entry · level 0 · unverified · cfb3154cdccd72da
  2. Embedder helper · level 1 · ran · 2def3130991aaf3c
  3. Head helper · level 1 · ran · b4868fbed7dce84a
  4. KeysValues helper · level 1 · ran · fba3ef4043569d6f
  5. LossWithIntermediateLosses helper · level 1 · ran · d7a1b7c2fadc881a
  6. Tokenizer model · level 0 · unverified · 79b0b0b918ab2ac4
  7. Transformer model · level 0 · unverified · d65bf756384c37cd
  8. TransformerConfig helper · level 1 · ran · 06c7be0a3042efdc
  9. WorldModelOutput helper · level 1 · ran · 419935433daf0f1b
  10. init_weights helper · level 0 · unverified · 7fc674de9a0a0a02
  11. Slicer helper · level 1 · ran · 3e962c6d7a040002
  12. KVCache helper · level 1 · ran · 93bd9c4b1f02db59
  13. Decoder helper · level 0 · unverified · 4e93d0d0e4839243
  14. Encoder helper · level 0 · unverified · aee0695eaeb766c4
  15. TokenizerEncoderOutput helper · level 1 · ran · bf4dc4c0382a6bd0
  16. Block helper · level 0 · unverified · ee23a284027c25be
  17. Cache helper · level 1 · ran · 256e6e8e5c0039da
  18. AttnBlock helper · level 1 · ran · 9d1f4ad302b969e4
  19. EncoderDecoderConfig helper · level 1 · ran · cd65aa1097b08332
  20. Normalize helper · level 1 · ran · our draft was wrong · 41ffde6c773af0d6
  21. ResnetBlock helper · level 0 · unverified · caffc15eb259ebb0
  22. Upsample helper · level 1 · ran · 53918c7471cbccdc
  23. nonlinearity helper · level 1 · ran · honoured contract · 3f6efbafceb71819
  24. Downsample helper · level 1 · ran · e8df5d8fcd9e9339
  25. SelfAttention helper · level 0 · unverified · f16fd90513419812
  26. AssignWithoutInplaceCheck helper · level 1 · ran · 7f939e4da5014501

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="2209.00588") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Atari Games 100kDeep Reinforcement LearningReinforcement LearningReinforcement Learning (RL)reinforcement-learning

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

Absolute Position EncodingsAdamAttentionAverage PoolingBPEBatch NormalizationConvolutionDense ConnectionsDropoutLabel SmoothingLayer NormalizationLinear LayerMonte-Carlo Tree SearchMuZeroMulti-Head AttentionPosition-Wise Feed-Forward LayerPrioritized Experience ReplayReLUResidual BlockResidual ConnectionSoftmaxTransformer

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