TTMM pre-trains many local LoRA experts on data clusters and merges the most relevant few at test time, approximating test-time training with a 100x speedup and near-TTT perplexity.
Mediator: Memory-efficient LLM Merging with Less Parameter Conflicts and Uncertainty Based Routing
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
Model merging aggregates Large Language Models (LLMs) finetuned on different tasks into a stronger one. However, parameter conflicts between models leads to performance degradation in averaging. While model routing addresses this issue by selecting individual models during inference, it imposes excessive storage and compute costs, and fails to leverage the common knowledge from different models. In this work, we observe that different layers exhibit varying levels of parameter conflicts. Building on this insight, we average layers with minimal parameter conflicts and use a novel task-level expert routing for layers with significant conflicts. To further reduce storage costs, inspired by task arithmetic sparsity, we decouple multiple fine-tuned experts into a dense expert and several sparse experts. Considering the out-of-distribution samples, we select and merge appropriate experts based on the task uncertainty of the input data. We conduct extensive experiments on both LLaMA and Qwen with varying parameter scales, and evaluate on real-world reasoning tasks. Results demonstrate that our method consistently achieves significant performance improvements while requiring less system cost.
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Local Mixtures of Experts: Essentially Free Test-Time Training via Model Merging
TTMM pre-trains many local LoRA experts on data clusters and merges the most relevant few at test time, approximating test-time training with a 100x speedup and near-TTT perplexity.