autoregressive model
PulseAugur coverage of autoregressive model — every cluster mentioning autoregressive model across labs, papers, and developer communities, ranked by signal.
4 day(s) with sentiment data
-
New framework GeoSim analyzes VLM representations for image restoration
Researchers have developed a new framework called GeoSim to analyze the internal representations of vision-language models (VLMs) for low-level image restoration tasks. The study investigates how different VLM architect…
-
AI Video Generation Better Aligns with Human Visual Cortex
Researchers have developed a new method to evaluate video generation models by comparing their internal representations to human visual cortex activity. They found that autoregressive (AR) video diffusion models, when g…
-
New SPRINT model generates recommendations in single step, boosting efficiency and accuracy
Researchers have introduced SPRINT, a novel single-step generative recommendation system that bypasses the token-by-token generation common in existing autoregressive and non-autoregressive models. By viewing item recom…
-
Diffusion language models achieve over 1,000 tokens/sec speed
Diffusion language models are emerging as a new paradigm, capable of generating text at speeds exceeding 1,000 tokens per second. This marks a significant shift from traditional autoregressive models, which generate tex…
-
AI startup VIDRAFT unveils AX-RAY to detect hidden causal leakage in models
VIDRAFT, an AI safety startup, has developed a new auditing framework called AX-RAY to detect causal leakage in hybrid sequence models. This leakage occurs when future token information improperly influences earlier pos…
-
MegaParts scales 3D object generation to 300 parts using token-efficient autoregressive modeling
Researchers have developed MegaParts, a novel framework for 3D object generation that significantly scales part-aware modeling. By employing token-efficient vector-quantized part tokens and structured autoregressive seq…
-
New LHSDet method detects high-resolution AI-generated images using VQA
Researchers have developed LHSDet, a new method for detecting high-resolution AI-generated images. This approach reframes the detection task as a visual question answering problem, utilizing a vision-language framework.…
-
Continuous Interaction Diffusion (CID) architecture enhances LLM tool use
Researchers have introduced Continuous Interaction Diffusion (CID), a novel architecture for diffusion language models (dLLMs) that enhances their ability to use external tools. Unlike autoregressive models that pause g…
-
New research explores transformers for modeling dynamical systems · 2 sources tracked
Two new arXiv papers explore the application of transformer models to understanding and predicting dynamical systems. The first paper analyzes the mechanistic properties of single-layer transformers, interpreting causal…
-
New HERO training method boosts long-horizon accuracy for neural operators
Researchers have introduced HERO (History-Enriched Rollout Training), a novel method designed to improve the long-horizon accuracy of autoregressive neural operators. This technique addresses the issue of error accumula…
-
Survey paper details advancements in text-based image editing
This survey paper provides a comprehensive review of Instruction-based Image Editing (IIE), a field that leverages text prompts to modify images. It categorizes IIE research across data construction, model architectures…
-
LeapTalk framework achieves real-time talking-head generation at 200 FPS
Researchers have developed LeapTalk, a novel framework designed to overcome the latency-quality trade-off in talking-head generation. This new approach enables stable and real-time video generation with a single forward…
-
New methods estimate distribution differences in autoregressive models
Researchers have developed new methods to estimate the total variation (TV) distance between distributions generated by autoregressive models. These methods address the challenge that different inference engines, even w…
-
New Generative Refinement Networks advance visual synthesis benchmarks
Researchers have introduced Generative Refinement Networks (GRN), a novel visual synthesis paradigm designed to overcome the computational inefficiencies of diffusion models and the limitations of autoregressive models.…
-
New framework improves autoregressive image generation quality
Researchers have developed a new framework called Information-Grounding Guidance (IGG) to improve the quality of images generated by autoregressive (AR) models. This method addresses the issue of information inconsisten…
-
Apple researchers advance diffusion language models with new decoding techniques
Apple's Machine Learning Research division has published several papers detailing advancements in diffusion language models (dLLMs). These models offer potential for faster inference compared to autoregressive models by…
-
New 7B Uniform Diffusion Language Model 'Sumi' Released, Alongside Diffusion Model Advancements
Researchers have introduced Sumi, a 7-billion parameter uniform diffusion language model (UDLM) pretrained from scratch on 1.5 trillion tokens. This open-source model demonstrates competitive performance against autoreg…
-
New Reflective Masking technique enables multi-turn reasoning in diffusion models
Researchers have introduced Reflective Masking (RM), a post-training technique that enables Mask Diffusion Models (MDMs) to perform multi-turn reasoning through iterative self-revision. Unlike autoregressive models that…
-
New Parallel Jacobi Decoding speeds up image generation models
Researchers have developed a new method called Parallel Jacobi Decoding (PJD) to speed up autoregressive image generation models. This technique expands draft tokens in a two-dimensional spatial domain, allowing for par…
-
New method enhances compositional generalization in autoregressive models
Researchers have developed a new method for composing autoregressive models, drawing inspiration from composition strategies used in diffusion models. This approach, based on a factorized-conditionals assumption, ensure…