PARAMΔ upcycles dense models to MoE for per-language experts and grafts post-training deltas to enable data-efficient language expansion while preserving original capabilities.
arXiv preprint arXiv:2502.07346 , year=
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
fields
cs.CL 3years
2026 3verdicts
UNVERDICTED 3representative citing papers
Systematic experiments demonstrate that multilingual coverage in LLM post-training improves results for all languages and tasks compared to English-only, with low-resource languages gaining most and zero-shot transfer emerging at high diversity.
Treating language as a latent variable via polyGRPO RL improves Qwen2.5-7B-Instruct by 6.72% on English reasoning benchmarks and 6.89% on multilingual ones, with cross-task gains on commonsense reasoning from math-only training.
citing papers explorer
-
A Data-Efficient Path to Multilingual LLMs: Language Expansion via Post-training PARAM$\Delta$ Integration into Upcycled MoE
PARAMΔ upcycles dense models to MoE for per-language experts and grafts post-training deltas to enable data-efficient language expansion while preserving original capabilities.
-
English is Not All You Need: Systematically Exploring the Role of Multilinguality in LLM Post-Training
Systematic experiments demonstrate that multilingual coverage in LLM post-training improves results for all languages and tasks compared to English-only, with low-resource languages gaining most and zero-shot transfer emerging at high diversity.
-
Language as a Latent Variable for Reasoning Optimization
Treating language as a latent variable via polyGRPO RL improves Qwen2.5-7B-Instruct by 6.72% on English reasoning benchmarks and 6.89% on multilingual ones, with cross-task gains on commonsense reasoning from math-only training.