SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.10815 · 2026

Why Does RL Generalize Better Than SFT? A Data-Centric Perspective on VLM Post-Training

Wei Li, Hangjie Yuan, Tao Feng, Yanan Sun, Aojun Lu

arXiv · PDF · Open in the Atlas

Code that ran

3 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 2 failures of our instrument; 1 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2602.10815.json · badge: /api/ran/2602.10815.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
byyx666/DC-SFT community
Papers with Code archive
3 ran · 1 no recorded run
FunctionStatusWhere it lives
LossType Ran
ran on a synthesized input · constructed an object; no result was computed
byyx666/DC-SFT/swift/plugin/loss.py
pointer only (licence: NONE) · get_code("ab5966728910f65d")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, byyx666/DC-SFT when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

ce_loss_func Ran
ran; our contract draft was wrong, not the code
byyx666/DC-SFT/swift/plugin/loss.py
pointer only (licence: NONE) · get_code("b40d46077f78e7de")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, byyx666/DC-SFT when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

loss_scale_func Ran
ran; our contract draft was wrong, not the code
byyx666/DC-SFT/swift/plugin/loss.py
pointer only (licence: NONE) · get_code("38b5df9f0cfa4632")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, byyx666/DC-SFT when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

register_loss_func Not yet run
unverified (harvested, no recorded run)
byyx666/DC-SFT/swift/plugin/loss.py
pointer only (licence: NONE) · get_code("5a3ac54293be7e23")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, byyx666/DC-SFT when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 3 the same month, 0 later) and 1 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
ce_loss_func
brick of this paper's assembly: loss
4 other papers
3 dated earlier, 1 the same month
2411.18145 (2024-11)
CHOICE: Benchmarking the Remote Sensing Capabilities of Large Vision-Language Models
2506.14674 (2025-06)
Recognition through Reasoning: Reinforcing Image Geo-localization with Large Vision-Language Models
2509.17589 (2025-09)
Table2LaTeX-RL: High-Fidelity LaTeX Code Generation from Table Images via Reinforced Multimodal Language Models
and 1 more not named here
loss_scale_func
brick of this paper's assembly: entry
3 other papers
2 dated earlier, 1 the same month
2506.14674 (2025-06)
Recognition through Reasoning: Reinforcing Image Geo-localization with Large Vision-Language Models
2509.17589 (2025-09)
Table2LaTeX-RL: High-Fidelity LaTeX Code Generation from Table Images via Reinforced Multimodal Language Models
2602.11909 (2026-02)
Echo: Towards Advanced Audio Comprehension via Audio-Interleaved Reasoning
register_loss_func
brick of this paper's assembly: helper
3 other papers
2 dated earlier, 1 the same month
2506.14674 (2025-06)
Recognition through Reasoning: Reinforcing Image Geo-localization with Large Vision-Language Models
2509.17589 (2025-09)
Table2LaTeX-RL: High-Fidelity LaTeX Code Generation from Table Images via Reinforced Multimodal Language Models
2602.11909 (2026-02)
Echo: Towards Advanced Audio Comprehension via Audio-Interleaved Reasoning

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

The adaptation of large-scale Vision-Language Models (VLMs) through post-training reveals a pronounced generalization gap: models fine-tuned with Reinforcement Learning (RL) consistently achieve superior out-of-distribution (OOD) performance compared to those trained with Supervised Fine-Tuning (SFT). This paper posits a data-centric explanation for this phenomenon, contending that RL's generalization advantage arises from an implicit data filtering mechanism that inherently prioritizes medium-difficulty training samples. To test this hypothesis, we systematically evaluate the OOD generalization of SFT models across training datasets of varying difficulty levels. Our results confirm that data difficulty is a critical factor, revealing that training on hard samples significantly degrades OOD performance. Motivated by this finding, we introduce Difficulty-Curated SFT (DC-SFT), a straightforward method that explicitly filters the training set based on sample difficulty. Experiments show that DC-SFT not only substantially enhances OOD generalization over standard SFT, but also surpasses the performance of RL-based training, all while providing greater stability and computational efficiency. This work offers a data-centric account of the OOD generalization gap in VLMs and establishes a more efficient pathway to achieving robust generalization. Code is available at https://github.com/byyx666/DC-SFT

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2602.10815")
get_code_for_paper("2602.10815")
have("2602.10815")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2602.10815.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2602.10815.svg)](https://syntology.ai/paper/2602.10815)

Connect an agent — have() is free.