arXiv:2607. 16062v1 Announce Type: cross Abstract: Model merging is promoted as a substitute for joint multi-task training, yet in the reinforcement-learning setting this substitution is essentially never tested against the baseline it claims to replace: methods merge independently released agents precisely because a joint model is unavailable.
arXiv:2606. 18627v1 Announce Type: new Abstract: Model merging has emerged as a training-free alternative to multi-task learning, aiming to combine multiple task-specific fine-tuned models into a single multi-task model.
The paper investigates three fusion paradigms—Merge, Mix RL, and multi‑teacher on‑policy distillation (MOPD)—for consolidating reinforcement learning with verifiable rewards (RLVR) across multiple domains. Experiments across model scales and a multi‑domain benchmark show that while overall performance differences are small, significant gaps can appear on specific tasks, and each method exhibits distinct training dynamics and constraints. Practical guidelines are offered: Merge for cheap fusion when experts exist, Mix RL for unified training with adjustable domain mixtures, and MOPD when preserving domain‑specific gains is paramount.
arXiv:2606. 19549v1 Announce Type: new Abstract: Low-rank adaptation (LoRA) makes it cheap to train many domain- and task-specific language model adapters, but whether two adapters can be merged is usually discovered only after both have been fully trained and evaluated.
arXiv:2610.03199v1 Announce Type: new Abstract: Large language models fine-tuned from a shared base can be merged by averaging their task vectors, but some merges collapse far below the base model, a...
ResMerge is a new framework for merging large language models trained via reinforcement learning. It separates each model’s task vector into a leading spectral head and a residual component, finding that both parts contain valuable behavior knowledge but behave differently during merging. The method builds a stable residual backbone using Spherical Residual Consensus Adaptation and then adds a lightweight head correction module that activates only when experts agree, leading to better preservation of expert capabilities compared to existing merging baselines.
The paper introduces DiGA, a Disentangled Geometry-Aware framework for merging pretrained models. DiGA orthogonally decomposes each task vector into components tied to distinct geometric attributes, aggregates these components independently, and then recombines them, thereby preserving each component’s geometric identity. Experiments across various models, tasks, and merging methods show that DiGA improves merged-model performance and reduces capability degradation.
The paper introduces a new multi‑task model‑merging framework that tackles task interference by projecting task vectors into a high‑dimensional sparse feature space using Sparse Autoencoders, enabling feature‑level disentanglement before fusion. It also proposes a lightweight Group‑Ranked Zeroth‑Order Optimizer to identify task‑critical layers for selective merging, reducing computational overhead. Experiments on Qwen2.5‑1.5B and Qwen2.5‑7B show consistent performance gains over several baselines across reasoning, code generation, instruction following, and general knowledge tasks, with a 2.78% improvement in a highly conflicting four‑task setting.
Mixture-Trained Merging (MTM) is a method for creating unified language models that combine multiple objectives—such as mathematics, code, instruction following, and controllable thinking—into a single parameter set. Instead of sequentially post‑training on each objective, MTM trains each branch on a mixture of objectives, ensuring that the branches remain compatible in weight space and can be merged without degrading performance. The approach iteratively refines merge coefficients using low‑cost evaluations and multi‑objective Bayesian optimization, outperforming naive merging and preserving distinct behaviors across domains.
arXiv:2607. 11997v1 Announce Type: cross Abstract: Multi-task model merging combines separately trained expert models into a single model that handles all tasks without co-training.
arXiv:2606. 18521v1 Announce Type: cross Abstract: Reinforcement Learning with Verifiable Reward (RLVR) has emerged as a powerful post-training paradigm that surpasses Supervised Fine-Tuning (SFT) in eliciting reasoning intelligence and resisting catastrophic forgetting.
arXiv:2607. 20561v1 Announce Type: new Abstract: LoRA adapters provide an efficient way to specialize a pretrained model for many downstream tasks, but deploying one adapter per task requires adapter storage and task selection at inference time.
CAMFT is a Conflict‑Aware Mergeable Fine‑Tuning method designed to make task adaptation efficient and merge‑aware for large language models. Unlike existing approaches that only resolve parameter conflicts after fine‑tuning, CAMFT shapes mergeability during training by guiding each task to update sparse coordinates with lower cross‑task conflict. Experiments show that CAMFT outperforms standard fine‑tuning baselines in multi‑task merging scenarios.