BidirLM: From Text to Omnimodal Bidirectional Encoders by Adapting and Composing Causal LLMs

• URL: http://arxiv.org/abs/2604.02045v1
• Date: Thu, 02 Apr 2026 13:48:35 GMT
• Title: BidirLM: From Text to Omnimodal Bidirectional Encoders by Adapting and Composing Causal LLMs
• Abstract summary: BidirLM is a family of five encoders that outperform alternatives on text, vision, and audio representation benchmarks.<n>We identify the key factors driving successful adaptation, highlighting the role of an often-omitted prior masking phase.<n>We augment our encoders by merging them with specialized causal models, seamlessly transferring modality- and domain-specific capabilities.
• Score: 11.874112583811593
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Transforming causal generative language models into bidirectional encoders offers a powerful alternative to BERT-style architectures. However, current approaches remain limited: they lack consensus on optimal training objectives, suffer from catastrophic forgetting at scale, and fail to flexibly integrate the vast ecosystem of specialized generative models. In this work, through systematic ablations on the Gemma3 and Qwen3 families, we identify the key factors driving successful adaptation, highlighting the critical role of an often-omitted prior masking phase. To scale this process without original pre-training data, we introduce a dual strategy combining linear weight merging with a lightweight multi-domain data mixture that mitigates catastrophic forgetting. Finally, we augment our encoders by merging them with specialized causal models, seamlessly transferring modality- and domain-specific capabilities. This open-source recipe, designed for any causal decoder LLM, yields BidirLM, a family of five encoders that outperform alternatives on text, vision, and audio representation benchmarks.

Related papers

• Context-Aware Cluster Decoding: Semantic Anchor-Driven Coherence in dMLLMs [7.894740505622926]
  Diffusion multimodal large language models (dMLLMs) produce long-form outputs marred by semantic drift and repetition.<n>We propose ours (textbfContext-textbfAware textbfCluster textbfDecoding), a training-free decoding method that scores each masked position by a multiplicative composite of softmax confidence and neighbor proximity.
  arXiv  Detail & Related papers  (2026-08-23T11:16:04Z)
• Beyond Text Conditioning: A Systematic Study of MLLM-DiT Fusion for Video Generation [50.92925099642139]
  Diffusion Transformers (DiTs) have become the dominant paradigm for high-fidelity video generation, yet their ability to perform high-level semantic planning remains limited.<n>We propose BiVidGen, a hybrid framework where an MLLM first generates semantic visual tokens and a DiT renders videos conditioned on both text and these tokens via multi-layer cross-attention.
  arXiv  Detail & Related papers  (2026-08-14T07:40:41Z)
• FLARE: Diffusion for Hybrid Language Model [72.60770374799634]
  FLARE is a systematic conversion framework for hybrid-attention large language models.<n>It combines a token-equal AR-and-diffusion objective, hardware-aware kernels, and unified inference, enabling one checkpoint to support both AR-style verified decoding and diffusion-style parallel denoising.<n>Our results suggest practical dLLMs are limited not only by decoding algorithms, but also by transfer data quality and the training inefficiency of current block-diffusion objectives.
  arXiv  Detail & Related papers  (2026-06-01T06:58:15Z)
• Neural Router: Semantic Content Matching for Agentic AI [41.39228253084888]
  Large language models (LLMs) can serve as the semantic-matching engine of a content-based publish/subscribe broker for agentic AI.<n> Framed as offline multi-label retrieval over three public datasets, our central contribution is a two-crossover cost-accuracy characterisation.
  arXiv  Detail & Related papers  (2026-05-25T10:58:53Z)
• Cheers: Decoupling Patch Details from Semantic Representations Enables Unified Multimodal Comprehension and Generation [66.53544128707817]
  Cheers is a unified multimodal model that decouples patch-level details from semantic representations.<n> Cheers matches or surpasses advanced UMMs in both visual understanding and generation.
  arXiv  Detail & Related papers  (2026-03-13T08:55:27Z)
• Hephaestus: Mixture Generative Modeling with Energy Guidance for Large-scale QoS Degradation [44.97875113025023]
  We study the Quality of Service Degradation (QoSD) problem, in which an adversary perturbs edge weights to degrade network performance.<n>No prior model directly tackles the RefineD problem under nonlinear edge-weight functions.<n>This work proposes PIMMA, a self-reinforcing framework that synthesizes feasible solutions in latent space.
  arXiv  Detail & Related papers  (2025-10-19T22:48:35Z)
• MOBIUS: Big-to-Mobile Universal Instance Segmentation via Multi-modal Bottleneck Fusion and Calibrated Decoder Pruning [91.90342432541138]
  Scaling up model size and training data has advanced foundation models for instance-level perception.<n>High computational cost limits adoption on resource-constrained platforms.<n>We introduce a new benchmark for efficient segmentation on both high-performance computing platforms and mobile devices.
  arXiv  Detail & Related papers  (2025-10-16T18:00:00Z)
• Leveraging Importance Sampling to Detach Alignment Modules from Large Language Models [48.15777554876988]
  Traditional alignment methods often require retraining large pretrained models.<n>We propose a novel textitResidual Alignment Model (textitRAM) that formalizes the alignment process as a type of importance sampling.<n>We develop a resampling algorithm with iterative token-level decoding to address the common first-token latency issue in comparable methods.
  arXiv  Detail & Related papers  (2025-05-26T08:53:02Z)
• Nexus-Gen: Unified Image Understanding, Generation, and Editing via Prefilled Autoregression in Shared Embedding Space [9.327655601475605]
  We propose Nexus-Gen, a novel architecture that unifies image understanding, generation, and editing tasks in a shared image embedding space.<n>To mitigate the severe error accumulation during autoregressive embedding prediction, we propose a novel prefilled autoregression strategy.<n>Nexus-Gen achieves state-of-the-art performance on the evaluation benchmarks spanning image understanding, generation and editing tasks.
  arXiv  Detail & Related papers  (2025-04-30T06:30:48Z)
• Encoder-Decoder Gemma: Improving the Quality-Efficiency Trade-Off via Adaptation [52.19855651708349]
  We study a novel problem: adapting decoder-only large language models to encoder-decoder models.<n>We argue that adaptation not only enables inheriting the capability of decoder-only LLMs but also reduces the demand for computation.<n>Under similar inference budget, encoder-decoder LLMs achieve comparable (often better) pretraining performance but substantially better finetuning performance than their decoder-only counterpart.
  arXiv  Detail & Related papers  (2025-04-08T17:13:41Z)
• Unbiased Max-Min Embedding Classification for Transductive Few-Shot Learning: Clustering and Classification Are All You Need [83.10178754323955]
  Few-shot learning enables models to generalize from only a few labeled examples.<n>We propose the Unbiased Max-Min Embedding Classification (UMMEC) Method, which addresses the key challenges in few-shot learning.<n>Our method significantly improves classification performance with minimal labeled data, advancing the state-of-the-art in annotatedL.
  arXiv  Detail & Related papers  (2025-03-28T07:23:07Z)
• Read-ME: Refactorizing LLMs as Router-Decoupled Mixture of Experts with System Co-Design [59.00758127310582]
  We propose a novel framework Read-ME that transforms pre-trained dense LLMs into smaller MoE models. Our approach employs activation sparsity to extract experts. Read-ME outperforms other popular open-source dense models of similar scales.
  arXiv  Detail & Related papers  (2024-10-24T19:48:51Z)
• Exploring the Role of Large Language Models in Prompt Encoding for Diffusion Models [42.891427362223176]
  Large language models (LLMs) based on decoder-only transformers have demonstrated superior text understanding capabilities.<n>We propose a novel framework to fully harness the capabilities of LLMs.<n>We further design an LLM-Infused Diffusion Transformer (LI-DiT) based on the framework.
  arXiv  Detail & Related papers  (2024-06-17T17:59:43Z)
• Large-scale Transfer Learning for Low-resource Spoken Language Understanding [31.013231069185387]
  We propose an attention-based Spoken Language Understanding model together with three encoder enhancement strategies to overcome data sparsity challenge. Cross-language transfer learning and multi-task strategies have been improved by up to 4:52% and 3:89% respectively, compared to the baseline.
  arXiv  Detail & Related papers  (2020-08-13T03:43:05Z)

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.