Scaling Linear Mode Connectivity and Merging to Billion Parameter Pretrained Transformers

• URL: http://arxiv.org/abs/2606.23607v1
• Date: Mon, 22 Jun 2026 17:08:28 GMT
• Title: Scaling Linear Mode Connectivity and Merging to Billion Parameter Pretrained Transformers
• Abstract summary: We propose a novel framework for enabling LMC-based model merging to pretrained transformers.<n>Our method applies properly parameterized functionality-preserving transformations and introduces a dual learning procedure in which both models jointly learn their corresponding solutions.<n> Empirically, we show that our approach near-zero loss barriers on WikiText for language models with medium-sized parameters, representing, to our knowledge, the first demonstration of near-barrier-free linear connectivity at this scale.
• Score: 42.7047638496247
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Linear mode connectivity (LMC) provides a promising foundation for understanding and merging independently trained neural networks, but existing methods typically optimize the interpolation path from only one model endpoint, limiting their scalability and effectiveness for large pretrained transformers. We propose a novel and scalable framework for enabling LMC-based model merging to {\em billion-parameter pretrained transformers}. Our method applies properly parameterized functionality-preserving weight transformations to align functionally equivalent solutions, and introduces a dual learning procedure in which both models jointly learn their corresponding transformations toward a shared linear interpolation path. This bidirectional optimization substantially reduces interpolation barriers and enables more reliable merging across large-scale architectures. Empirically, we show that our approach achieves near-zero loss barriers on WikiText for language models with medium-sized parameters, representing, to our knowledge, the first demonstration of near-barrier-free linear connectivity at this scale. In the vision domain, ViT-L maintains above 69\% ImageNet top-1 accuracy throughout the interpolation path, while modern billion-parameter LLMs exhibit only small loss barriers. These results suggest that properly resolving parameter symmetries enables large pretrained Transformers to be connected and merged through simple linear paths with substantially improved interpolation performance. Code: https://github.com/VILA-Lab/Dual-Learned-Matching .

Related papers

• Rethinking Heterogeneous LLM Merging: A Weighted Model Averaging Perspective [41.30735431552607]
  heterogeneous fusion methods typically introduce adapters, learned latent spaces, routing, or feature alignment.<n>We revisit this counterintuitive question through training-free dimensional adaptation followed by ratio-controlled ratios.<n>Results show that simple parameter averaging, when paired with lightweight dimensional adaptation and carefully controlled ratios, is a surprisingly strong baseline for heterogeneous merging.
  arXiv  Detail & Related papers  (2026-07-20T14:58:53Z)
• cMoLLM at Scale: Horizontal Scaling Laws for Mixture-of-LLMs [3.50654760771471]
  cMoLLM is a convolutionally gated mixture-of-LLMs that routes over end-to-end streams through fully differentiable dynamic convolution.<n>In GPT-2-style models trained on FineWeb, cMoLLM improves language modeling perplexity and downstream GLUE and SQuAD accuracy under matched compute.
  arXiv  Detail & Related papers  (2026-06-06T13:51:15Z)
• Hybrid Dual-Path Linear Transformations for Efficient Transformer Architectures [0.0]
  We introduce the Hybrid Dual-Path Linear (HDPL) operator, which decomposes the affine transformation into two topologically distinct pathways.<n> Experiments on the FineWeb-Edu dataset demonstrate that the HDPL architecture outperforms a standard Llama-style baseline.<n>We discuss how the explicit materialization of a probabilistic latent space within the Transformer backbone serves as a vital architectural affordance.
  arXiv  Detail & Related papers  (2026-02-05T20:16:10Z)
• Merging Beyond: Streaming LLM Updates via Activation-Guided Rotations [55.047454145941366]
  Streaming Merging is an innovative model updating paradigm that conceptualizes merging as an iterative optimization process.<n> ARM is a strategy designed to approximate gradient descent dynamics.<n> ARM requires only early SFT checkpoints and, through iterative merging, surpasses the fully converged SFT model.
  arXiv  Detail & Related papers  (2026-02-03T08:15:57Z)
• Improving Recursive Transformers with Mixture of LoRAs [2.672804414228544]
  Mixture of LoRAs (MoL) inserts Low-Rank Adaptation (LoRA) experts inside a shared feed-forward network (FFN)<n>MoL enables token-conditional weight-space modulation of the shared FFN without untying backbone parameters.<n>ModernALBERT achieves state-of-the-art performance among compact models.
  arXiv  Detail & Related papers  (2025-12-14T23:39:30Z)
• Transformer-based Scalable Beamforming Optimization via Deep Residual Learning [12.79709425087431]
  unsupervised deep learning framework for downlink beamforming in large-scale MU-MISO channels.<n>Model is trained offline, allowing real-time inference through lightweight feedforward computations in dynamic communication environments.
  arXiv  Detail & Related papers  (2025-10-15T01:43:51Z)
• CollaPipe: Adaptive Segment-Optimized Pipeline Parallelism for Collaborative LLM Training in Heterogeneous Edge Networks [57.95170323315603]
  We introduce CollaPipe, a distributed learning framework that integrates collaborative pipeline parallelism with federated aggregation to support self-evolving networks.<n>In CollaPipe, the encoder part is adaptively partitioned into variable-sized segments and deployed across mobile devices for pipeline-parallel training, while the decoder is deployed on edge servers to handle generative tasks.<n>To enhance training efficiency, we formulate a joint optimization problem that adaptively allocates model segments, micro-batches, bandwidth, and transmission power.
  arXiv  Detail & Related papers  (2025-09-24T07:54:01Z)
• CARE Transformer: Mobile-Friendly Linear Visual Transformer via Decoupled Dual Interaction [77.8576094863446]
  We propose a new detextbfCoupled dutextbfAl-interactive lineatextbfR atttextbfEntion (CARE) mechanism. We first propose an asymmetrical feature decoupling strategy that asymmetrically decouples the learning process for local inductive bias and long-range dependencies. By adopting a decoupled learning way and fully exploiting complementarity across features, our method can achieve both high efficiency and accuracy.
  arXiv  Detail & Related papers  (2024-11-25T07:56:13Z)
• Unifying Dimensions: A Linear Adaptive Approach to Lightweight Image Super-Resolution [6.857919231112562]
  Window-based transformers have demonstrated outstanding performance in super-resolution tasks. They exhibit higher computational complexity and inference latency than convolutional neural networks. We construct a convolution-based Transformer framework named the linear adaptive mixer network (LAMNet)
  arXiv  Detail & Related papers  (2024-09-26T07:24:09Z)
• SHERL: Synthesizing High Accuracy and Efficient Memory for Resource-Limited Transfer Learning [63.93193829913252]
  We propose an innovative METL strategy called SHERL for resource-limited scenarios. In the early route, intermediate outputs are consolidated via an anti-redundancy operation. In the late route, utilizing minimal late pre-trained layers could alleviate the peak demand on memory overhead.
  arXiv  Detail & Related papers  (2024-07-10T10:22:35Z)
• RWKV: Reinventing RNNs for the Transformer Era [54.716108899349614]
  We propose a novel model architecture that combines the efficient parallelizable training of transformers with the efficient inference of RNNs. We scale our models as large as 14 billion parameters, by far the largest dense RNN ever trained, and find RWKV performs on par with similarly sized Transformers.
  arXiv  Detail & Related papers  (2023-05-22T13:57:41Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.