Pith Number

pith:VS3FT6ZZ

pith:2026:VS3FT6ZZ2KRY7TWNPVLLWUVKXP

not attested not anchored not stored refs resolved

Dynamic Latent Routing

Amir Abdullah, Fangyuan Yu, Xin Su

Dynamic Latent Routing recovers globally optimal policies by composing learned sub-policies to improve low-data language model fine-tuning.

arxiv:2605.14323 v1 · 2026-05-14 · cs.LG · cs.AI · cs.CL

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{VS3FT6ZZ2KRY7TWNPVLLWUVKXP}

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

In low-data fine-tuning settings, DLR matches or outperforms supervised fine-tuning across four datasets and six models, achieving a mean gain of +6.6 percentage points, while prior discrete-latent baselines consistently underperform SFT.

C2weakest assumption

That the optimality guarantees and search principle from General Dijkstra Search in MDPs transfer effectively to the non-stationary, high-dimensional setting of language model post-training without introducing hidden biases or optimization instabilities.

C3one line summary

Dynamic Latent Routing jointly learns discrete latent codes, routing policies, and model parameters via dynamic search to match or exceed supervised fine-tuning by 6.6 points on average in low-data settings across four datasets and six models.

References

70 extracted · 70 resolved · 4 Pith anchors

[1] Hindsight experience replay 2017

[2] Introducing Claude Opus 4.7 2026

[3] The option-critic architecture 2017

[4] Successor features for transfer in reinforcement learning 2017

[5] Probing classifiers: Promises, shortcomings, and advances.Computational Linguistics, 48(1):207–219, 2022 2022

Formal links

2 machine-checked theorem links

Receipt and verification

First computed	2026-05-17T23:39:09.808381Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

acb659fb39d2a38fcecd7d56bb52aabbd5954034aff8d68f28aefd97cfaf27d8

Aliases

arxiv: 2605.14323 · arxiv_version: 2605.14323v1 · doi: 10.48550/arxiv.2605.14323 · pith_short_12: VS3FT6ZZ2KRY · pith_short_16: VS3FT6ZZ2KRY7TWN · pith_short_8: VS3FT6ZZ

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/VS3FT6ZZ2KRY7TWNPVLLWUVKXP \
  | 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: acb659fb39d2a38fcecd7d56bb52aabbd5954034aff8d68f28aefd97cfaf27d8

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "07849c9d1a4bf545b3e2a82e4b983491ba8c311b3d8a1b2b18df2693d350a8f2",
    "cross_cats_sorted": [
      "cs.AI",
      "cs.CL"
    ],
    "license": "http://creativecommons.org/licenses/by-sa/4.0/",
    "primary_cat": "cs.LG",
    "submitted_at": "2026-05-14T03:35:46Z",
    "title_canon_sha256": "053ff61afba5cb9c959c212834e1d656d977b42a269996ce53447947c04bb37a"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.14323",
    "kind": "arxiv",
    "version": 1
  }
}