Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

OptiMer: Optimal Distribution Vector Merging Is Better than Data Mixing for Continual Pre-Training

About

Continual pre-training is widely used to adapt LLMs to target languages and domains, yet the mixture ratio of training data remains a sensitive hyperparameter that is expensive to tune: they must be fixed before training begins, and a suboptimal choice can waste weeks of compute. In this work, we propose OptiMer, which decouples ratio selection from training: we train one CPT model per dataset, extract each model's distribution vector, which represents the parameter shift induced by that dataset, and search for optimal composition weights post-hoc via Bayesian optimization. Experiments on Gemma 3 27B across languages (Japanese, Chinese) and domains (Math, Code) show that OptiMer consistently outperforms data mixture and model averaging baselines with 15-35 times lower search cost. Key findings reveal that 1) the optimized weights can be interpreted as data mixture ratios, and retraining with these ratios improves data mixture CPT, and 2) the same vector pool can be re-optimized for a given objective without any retraining, producing target-tailored models on demand. Our work establishes that data mixture ratio selection, traditionally a pre-training decision, can be reformulated as a post-hoc optimization over distribution vectors, offering a more flexible paradigm for continual pre-training.

Haiyue Song, Masao Utiyama• 2026

Related benchmarks

TaskDatasetResultRank
Chinese Language UnderstandingC-Eval
Accuracy63.52
56
Question AnsweringTriviaQA (TQA)--
56
Mathematical ReasoningGSM8K
Accuracy (Acc)84.38
42
Language UnderstandingMMLU
MMLU Score73.89
12
Commonsense ReasoningHSWAG
HSwag Accuracy82.52
9
Code GenerationHE
HE Score77.44
4
Code GenerationMBPP
MBPP Score70.2
4
General Model CapabilityTable 3 Evaluation Suite
Average Score70.94
4
Japanese Language UnderstandingJA LB
JA LB Score74.4
4
Low-Resource Multilingual UnderstandingASEAN Lang
Average Score (11 Tasks)68.38
4
Showing 10 of 11 rows

Other info

GitHub

Follow for update