Papers › Pushdown Layers: Encoding Recursive Structure in Transformer Language Models

Pushdown Layers: Encoding Recursive Structure in Transformer Language Models

29 Oct 2023arXiv:2310.19089archive 2025-07-28

Shikhar Murty, Pratyusha Sharma, Jacob Andreas, Christopher D. Manning

Recursion is a prominent feature of human language, and fundamentally challenging for self-attention due to the lack of an explicit recursive-state tracking mechanism. Consequently, Transformer language models poorly capture long-tail recursive structure and exhibit sample-inefficient syntactic generalization. This work introduces Pushdown Layers, a new self-attention layer that models recursive state via a stack tape that tracks estimated depths of every token in an incremental parse of the observed prefix. Transformer LMs with Pushdown Layers are syntactic language models that autoregressively and synchronously update this stack tape as they predict new tokens, in turn using the stack tape to softly modulate attention over tokens -- for instance, learning to "skip" over closed constituents. When trained on a corpus of strings annotated with silver constituency parses, Transformers equipped with Pushdown Layers achieve dramatically better and 3-5x more sample-efficient syntactic generalization, while maintaining similar perplexities. Pushdown Layers are a drop-in replacement for standard self-attention. We illustrate this by finetuning GPT2-medium with Pushdown Layers on an automatically parsed WikiText-103, leading to improvements on several GLUE text classification tasks.

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 23 code samples harvested from 1 repository linked to this paper: 14 ran (of which 8 constructed an object rather than computing a result; 14 with no instrument failure: 1 honoured, 0 violated, 13 with no contract checked; 0 where Syntology's instrument failed) · 9 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code; here there are none.

By repository: official repository: 23 samples from 1 repository, 14 ran (14 with no instrument failure; 0 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 2026-09-22 and 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

murtyshikhar/pushdown-layers officialmentioned in paperpytorchSyntology: 23 samples, 14 ran (0 instrument-failed) · 9 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

23 samples harvested: 14 ran (of which 8 constructed an object rather than computing a result; 14 with no instrument failure: 1 honoured, 0 violated, 13 with no contract checked; 0 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
13ran
9unverified

Licence: 23 of the 23 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 murtyshikhar/pushdown-layers. “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 3 samples 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.

murtyshikhar/pushdown-layers official repository 23 samples, 14 ran (0 instrument-failed) · 9 unverified
AttentionMask models/pushdown_transformer_lm.py ran · metamorphic tier: deterministic no licence file found · pointer only · 551d6401ece551f3 · report fixture: returned AttentionMask · level recorded 2026-09-22
DotDict models/pushdown_transformer_lm.py ran · metamorphic tier: deterministic no licence file found · pointer only · a7a732117e3ad716 · report fixture: returned DotDict · level recorded 2026-09-22
MultiHeadAttentionBase models/pushdown_transformer_lm.py ran · metamorphic tier: deterministic no licence file found · pointer only · 6ea6715c721ed9a7 · report fixture: returned MultiHeadAttentionBase · level recorded 2026-09-22
PositionalEncoding in 126 papers
census: 13 implementations from 9 papers, 9 distinct outputs across 5 buckets, one shared input per bucket
models/pushdown_transformer_lm.py ran · metamorphic tier: deterministic no licence file found · pointer only · 545b1940b792fa22 · report fixture: returned PositionalEncoding · level recorded 2026-09-22
RelativeAttentionBase models/pushdown_transformer_lm.py ran · metamorphic tier: deterministic no licence file found · pointer only · 59316687afde843b · report fixture: returned RelativeAttentionBase · level recorded 2026-09-22
TiedEmbedding models/pushdown_transformer_lm.py ran · metamorphic tier: deterministic fingerprinted no licence file found · pointer only · 5612e13b2384e5b0 · report fixture: returned TiedEmbedding · level recorded 2026-09-22
TransformerEncoder in 96 papers models/pushdown_transformer_lm.py ran · metamorphic tier: deterministic no licence file found · pointer only · 217b95213cfe1f2b · report fixture: returned TransformerEncoder · level recorded 2026-09-22
TransformerResult models/pushdown_transformer_lm.py ran · metamorphic tier: deterministic no licence file found · pointer only · cd78addc129539b3 · report fixture: returned TransformerResult · level recorded 2026-09-22
compute_per_token_logprob pushdown_util.py ran no licence file found · pointer only · 70622ae82f44b450 · report fixture: signature lm:callable/0/?, str_logits:Tensor/3/float, inputs:Tensor/2/int, input_lens:Tensor/1/int · returned list · level recorded 2026-09-24
compute_stack_logprobs pushdown_util.py ran no licence file found · pointer only · dcf24b95d9f14aa9 · report fixture: signature lm:callable/0/?, attachment_logits:list/1/Tensor, stack_labels:list/1/list, input_lens:list/1/int · returned list · level recorded 2026-09-24
get_parsing_accuracy eval_utils/eval_pushdown_model.py ran no licence file found · pointer only · 45d4819a42863712 · report fixture: signature predicted_parses:list/1/tuple, gold_parses:list/1/tuple · returned dict · level recorded 2026-09-24
logsumexp in 55 papers
census: 8 implementations from 12 papers, 4 distinct outputs across 2 buckets, one shared input per bucket
beam_search_util.py ran fingerprinted no licence file found · pointer only · 6b331f505ce51944 · report fixture: signature x:ndarray/1/float · returned float64 · level recorded 2026-09-24
sinusoidal_pos_embedding models/pushdown_transformer_lm.py ran · honoured contract no licence file found · pointer only · f24ed9058bec8d35 · report fixture: signature d_model:int/0/int, max_len:int/0/int, pos_offset:int/0/int, device:NoneType/0/? · returned Tensor · level recorded 2026-09-22
sinusoidal_pos_embedding layers/positional_encoding.py ran no licence file found · pointer only · d6b9018a0880eafd · report fixture: signature d_model:int/0/int, max_len:int/0/int, pos_offset:int/0/int, device:NoneType/0/? · returned Tensor · level recorded 2026-09-24
PushdownAttention models/pushdown_transformer_lm.py unverified no licence file found · pointer only · 040a0db1fa164ee2 · report
PushdownLM models/pushdown_transformer_lm.py unverified no licence file found · pointer only · 1d6a2fb2322500d7 · report
StackPredictor models/pushdown_transformer_lm.py unverified no licence file found · pointer only · ea75be5e2691ee1e · report
Transformer in 173 papers models/pushdown_transformer_lm.py unverified no licence file found · pointer only · e1e85555ec42de79 · report
TransformerEncoderLayer in 90 papers models/pushdown_transformer_lm.py unverified no licence file found · pointer only · cd05f27177c97848 · report
TransformerEncoderWithLayer models/pushdown_transformer_lm.py unverified no licence file found · pointer only · 5d6f6fc2cd657722 · report
TransformerEncoderWithLayer layers/transformer/transformer.py unverified no licence file found · pointer only · 9f7833997a6a7918 · report
TransformerLM models/pushdown_transformer_lm.py unverified no licence file found · pointer only · 199613ff3592d354 · 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 Pushdown Layers (graph method: Pushdown Layers). 16 bricks; individually on a fixture: 9 ran (of which 8 constructed an object rather than computing a result; 9 with no instrument failure: 1 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 7 unverified. The assembly has not been run end-to-end.

  1. PushdownLM entry · level 0 · unverified · 1d6a2fb2322500d7
  2. AttentionMask helper · level 1 · ran · 551d6401ece551f3
  3. StackPredictor model · level 0 · unverified · ea75be5e2691ee1e
  4. Transformer helper · level 0 · unverified · e1e85555ec42de79
  5. TransformerLM helper · level 0 · unverified · 199613ff3592d354
  6. TransformerResult helper · level 1 · ran · cd78addc129539b3
  7. PushdownAttention model · level 0 · unverified · 040a0db1fa164ee2
  8. TransformerEncoderWithLayer helper · level 0 · unverified · 5d6f6fc2cd657722
  9. PositionalEncoding helper · level 1 · ran · 545b1940b792fa22
  10. TiedEmbedding helper · level 1 · ran · 5612e13b2384e5b0
  11. DotDict helper · level 1 · ran · a7a732117e3ad716
  12. RelativeAttentionBase helper · level 1 · ran · 59316687afde843b
  13. sinusoidal_pos_embedding helper · level 1 · ran · honoured contract · f24ed9058bec8d35
  14. TransformerEncoder helper · level 1 · ran · 217b95213cfe1f2b
  15. TransformerEncoderLayer helper · level 0 · unverified · cd05f27177c97848
  16. MultiHeadAttentionBase helper · level 1 · ran · 6ea6715c721ed9a7

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="2310.19089") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Text Classificationtext-classification

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

Absolute Position EncodingsAdamAttentionBPEDense ConnectionsDropoutLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual 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