Metraipolis › archive › dc_e99df269d4c4

Towards Scalable RLVR: Multimodal Instruction Following Data Synthesis and Distillation — arXiv:2609.16059

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.16059
License (as archived)
CC0-1.0
Archived
2026-10-15T19:19:05.467Z
SHA-256 of the archived text
f977a47cde968e13f44db0e319e93f952e88c0a3b76f9c3b2519049c5d514951
Stable URL of this page
https://metraipolis.com/documento/dc_e99df269d4c4

How to cite

Towards Scalable RLVR: Multimodal Instruction Following Data Synthesis and Distillation — arXiv:2609.16059 — arXiv (en), https://arxiv.org/abs/2609.16059 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e99df269d4c4 (Metraipolis archive dc_e99df269d4c4, sha256 f977a47cde968e13f44db0e319e93f952e88c0a3b76f9c3b2519049c5d514951)
BibTeX
@misc{dce99df269d4c4,
  title = {Towards Scalable RLVR: Multimodal Instruction Following Data Synthesis and Distillation — arXiv:2609.16059},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.16059},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e99df269d4c4; sha256 f977a47cde968e13f44db0e319e93f952e88c0a3b76f9c3b2519049c5d514951},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.16059. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Multimodal instruction following (MMIF) is crucial for building generalist agents. However, current training paradigms rely heavily on Supervised Fine-Tuning (SFT), which often leads to surface-level pattern matching and degrades general capabilities. While Reinforcement Learning with Verifiable Rewards (RLVR) offers a promising alternative, its scalability in MMIF is severely bottlenecked by the scarcity of high-quality, RL-ready multimodal data. To bridge this gap, we present MIFS (\textbf{M}ultimodal \textbf{I}nstruction \textbf{F}ollowing \textbf{S}ynthesis), a systematic pipeline designed to generate RL-ready multimodal data. Specifically, MIFS introduces a generative constraint protocol to synthesize diverse raw samples, followed by a learnability-aware distillation mechanism that filters data based on RL training dynamics to ensure stable policy optimization. Furthermore, a code-based verifier provides high-precision reward signals for policy learning. The resulting dataset comprises 90k samples across 8 constraint categories and 14 task domains. Empirical evaluations demonstrate that MIFS-trained MLLMs achieve an average improvement of 8.13\% on four MMIF benchmarks and a 3$\times$ faster training convergence compared to using raw data. Crucially, our approach mitigates the generalization trade-offs typical of SFT, preserving core visual capabilities while significantly boosting instruction-following precision.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-16 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.16059 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-15T18:32:32.794Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-16?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).