pith. sign in
Pith Number

pith:H2M3KJ73

pith:2026:H2M3KJ733MRYFZ3A6W24Y44MXL
not attested not anchored not stored refs resolved

Lost in Decoding? Reproducing and Stress-Testing the Look-Ahead Prior in Generative Retrieval

Kidist Amde Mekonnen, Maarten de Rijke, Simon Lupart, Yongkang Li, Yubao Tang

PAG's planning signal in generative retrieval collapses under intent-preserving typos.

arxiv:2604.23396 v1 · 2026-04-25 · cs.IR · cs.AI · cs.CL · cs.LG

Add to your LaTeX paper
\usepackage{pith}
\pithnumber{H2M3KJ733MRYFZ3A6W24Y44MXL}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp
2 Internet Archive
3 Author claim open · sign in to claim
4 Citations open
5 Replications open
Portable graph bundle live · download bundle · merged state
The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

PAG's planning signal is brittle under lexical surface-form variation: intent-preserving typos can trigger plan collapse, where the planned candidate pool shifts enough that the look-ahead bonus provides little useful guidance, effectively reverting decoding toward weaker unguided search.

C2weakest assumption

That the plan drift diagnostics and cross-lingual tests isolate the stability of the planning signal without being confounded by specific beam sizes, trie construction details, or unstated differences between the released checkpoint and the original training run.

C3one line summary

Reproduction confirms PAG boosts generative retrieval effectiveness, but its look-ahead planning signal collapses under intent-preserving typos and query mismatches, reverting performance to unguided decoding.

References

44 extracted · 30 resolved · 2 Pith anchors

[1] Michele Bevilacqua, Marco Maru, and Fabio Petroni. 2022. Autoregressive Search Engines: Generating Substrings as Document Identifiers. InAdvances in Neural Information Processing Systems (NeurIPS) 2022
[2] arXiv preprint arXiv:2108.13897 , year= 2021
[3] Steven Dong, Yubao Tang, and Maarten de Rijke. 2026. Multi-Step Semantic Rea- soning in Generative Retrieval. InEuropean Conference on Information Retrieval. Springer, 273–281 2026
[4] Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Sid- dharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaud- hary, Naman Goyal, Tom Birch, Vitaliy Liptchinsk 2021
[5] Tim Hagen, Harrisen Scells, and Martin Potthast. 2024. Revisiting Query Variation Robustness of Transformer Models. InFindings of the Association for Computa- tional Linguistics: EMNLP 2024, Yaser Al- 2024 · doi:10.18653/v1/2024.findings-emnlp.248
Receipt and verification
First computed 2026-05-26T01:03:31.176657Z
Builder pith-number-builder-2026-05-17-v1
Signature Pith Ed25519 (pith-v1-2026-05) · public key
Schema pith-number/v1.0

Canonical hash

3e99b527fbdb2382e760f5b5cc738cbad9069278ec43b042fd2ce00c565c1087

Aliases

arxiv: 2604.23396 · arxiv_version: 2604.23396v1 · doi: 10.48550/arxiv.2604.23396 · pith_short_12: H2M3KJ733MRY · pith_short_16: H2M3KJ733MRYFZ3A · pith_short_8: H2M3KJ73
Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/H2M3KJ733MRYFZ3A6W24Y44MXL \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 3e99b527fbdb2382e760f5b5cc738cbad9069278ec43b042fd2ce00c565c1087
Canonical record JSON
{
  "metadata": {
    "abstract_canon_sha256": "ca009d26a660ddfdc83d2ad226545f6b02427f8d17ebd14270feb38aeadff4c6",
    "cross_cats_sorted": [
      "cs.AI",
      "cs.CL",
      "cs.LG"
    ],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.IR",
    "submitted_at": "2026-04-25T17:58:15Z",
    "title_canon_sha256": "b2d8b74091a1eda5fbcbf95ff63dbb39e8bdbcfe97af4499246ca91f5954cbbe"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2604.23396",
    "kind": "arxiv",
    "version": 1
  }
}