Active lifting of data distributions via latent-sequence sampling and max-rate uncertainty reduction formally derives slow-thinking LLMs and places them on representation and sampler hierarchies that can be climbed.
hub Tool reference
Neural discrete representation learning.Advances in neural information processing systems, 30
Tool reference. 83% of classified Pith citations use this work as a method, library, or software dependency, not as a substantive claim.
23
Pith papers citing it
Method reference
83% of classified citations
hub tools
citation-role summary
method 5
background 1
citation-polarity summary
representative citing papers
GEAR jointly trains VQ tokenizer and AR generator end-to-end via dual hard/soft read-out and representation alignment, achieving up to 10x faster ImageNet gFID convergence than LlamaGen-REPA while generalizing across quantizers and to text-to-image.
AffectCodec applies block-diagonal projections in residual FSQ to explicitly allocate bits to emotion and acoustic subspaces, combined with emotion conditioning, yielding better emotion preservation at low bitrates with competitive acoustic quality.
The paper identifies distinct failure mechanisms: excessive posterior-prior regularization erases mode information in latent policies, while smooth base-to-action maps limit mode coverage in generative policies.
Vector quantization induces a structured partition of the representation space for composing heterogeneous multiclass calibration maps via shared codeword-dependent Dirichlet factors.
ToBAC is the first backdoor attack on unified autoregressive models, using data or model poisoning to make triggers elicit cross-modal malicious behavior in text and image generation.
RotVLA models latent actions as continuous SO(n) rotations with triplet-frame supervision and flow-matching to reach 98.2% success on LIBERO and 89.6%/88.5% on RoboTwin2.0 using a 1.7B-parameter model.
Spatial Gram Alignment aligns internal self-similarities of LDM features with foundation priors to reconcile global structure and fine details in ultra-high-resolution text-to-image synthesis.
CPC-VAR adds Gradient-based Concept Neuron Selection for continual single-concept learning and a context-aware multi-branch composition strategy to reduce forgetting and entanglement in VAR-based personalized image generation.
InsightTok improves text and face fidelity in discrete image tokenization via content-aware perceptual losses, with gains transferring to autoregressive generation.
ALAM introduces algebraic consistency regularization on latent action transitions from videos, raising VLA success rates from 47.9% to 85.0% on MetaWorld MT50 and 94.1% to 98.1% on LIBERO.
Yeti is a compact tokenizer for protein structures that delivers strong codebook use, token diversity, and reconstruction while enabling from-scratch multimodal generation of plausible sequences and structures with 10x fewer parameters than ESM3.
SDFlow learns a global transport map via similarity-driven flow matching in VQ latent space, using low-rank manifold decomposition and a categorical posterior to handle discreteness, yielding SOTA long-horizon performance and inference speedups.
CapsID uses probabilistic capsule routing and confidence-based termination to generate variable-length semantic IDs, improving recall by 9.6% over strong baselines with half the latency of dual-representation systems.
FaithfulFaces introduces a pose-faithful identity aligner with a shared dictionary and invariance constraint to maintain facial identity in text-to-video generation under large pose changes and occlusions.
R&B-EnCoRe uses self-supervised importance-weighted variational inference to distill action-predictive reasoning datasets that improve VLA performance on manipulation, navigation, and driving tasks without external verifiers.
NegBio-VAE introduces negative binomial latents with dispersion to handle overdispersion in discrete VAE models, yielding better reconstruction, generation, and downstream representations than Poisson VAE baselines.
IP-Adapter adds effective image prompting to text-to-image diffusion models using a lightweight decoupled cross-attention adapter that works alongside text prompts and other controls.
Rosetta proposes a composable multimodal pretraining method with MAOP to prevent catastrophic forgetting when expanding modalities beyond standard MoE and MoT approaches.
AID-VAR attaches an adversarial discriminator and lightweight guidance injector to frozen VAR backbones to diagnose and correct fidelity gaps across scales, reporting 16% FID gains with 3% added parameters.
TriProRep pretrains on three aligned protein views and improves results on homodimer co-folding and related structure tasks in the new RepSP benchmark.
VQ-VAE concept learning enables controllable recombination of crystal motifs to generate structures with reported gains in validity-stability-uniqueness-novelty metrics on MP-20 and Alex-MP-20.
The survey frames VLA models as pipelines that generate progressively grounded action tokens and classifies those tokens into eight types to guide future development.
citing papers explorer
-
A First-Principles Theory of Slow Thinking and Active Perception
Active lifting of data distributions via latent-sequence sampling and max-rate uncertainty reduction formally derives slow-thinking LLMs and places them on representation and sampler hierarchies that can be climbed.
-
GEAR: Guided End-to-End AutoRegression for Image Synthesis
GEAR jointly trains VQ tokenizer and AR generator end-to-end via dual hard/soft read-out and representation alignment, achieving up to 10x faster ImageNet gFID convergence than LlamaGen-REPA while generalizing across quantizers and to text-to-image.
-
AffectCodec: Emotion-Preserving Neural Speech Codec with Block-Diagonal Residual FSQ
AffectCodec applies block-diagonal projections in residual FSQ to explicitly allocate bits to emotion and acoustic subspaces, combined with emotion conditioning, yielding better emotion preservation at low bitrates with competitive acoustic quality.
-
Understanding Multimodal Failure in Action-Chunking Behavioral Cloning
The paper identifies distinct failure mechanisms: excessive posterior-prior regularization erases mode information in latent policies, while smooth base-to-action maps limit mode coverage in generative policies.
-
Divide et Calibra: Multiclass Local Calibration via Vector Quantization
Vector quantization induces a structured partition of the representation space for composing heterogeneous multiclass calibration maps via shared codeword-dependent Dirichlet factors.
-
Token by Token, Compromised: Backdoor Vulnerabilities in Unified Autoregressive Models
ToBAC is the first backdoor attack on unified autoregressive models, using data or model poisoning to make triggers elicit cross-modal malicious behavior in text and image generation.
-
RotVLA: Rotational Latent Action for Vision-Language-Action Model
RotVLA models latent actions as continuous SO(n) rotations with triplet-frame supervision and flow-matching to reach 98.2% success on LIBERO and 89.6%/88.5% on RoboTwin2.0 using a 1.7B-parameter model.
-
Spatial Gram Alignment for Ultra-High-Resolution Image Synthesis
Spatial Gram Alignment aligns internal self-similarities of LDM features with foundation priors to reconcile global structure and fine details in ultra-high-resolution text-to-image synthesis.
-
CPC-VAR:Continual Personalized and Compositional Generation in Visual Autoregressive Models
CPC-VAR adds Gradient-based Concept Neuron Selection for continual single-concept learning and a context-aware multi-branch composition strategy to reduce forgetting and entanglement in VAR-based personalized image generation.
-
InsightTok: Improving Text and Face Fidelity in Discrete Tokenization for Autoregressive Image Generation
InsightTok improves text and face fidelity in discrete image tokenization via content-aware perceptual losses, with gains transferring to autoregressive generation.
-
ALAM: Algebraically Consistent Latent Action Model for Vision-Language-Action Models
ALAM introduces algebraic consistency regularization on latent action transitions from videos, raising VLA success rates from 47.9% to 85.0% on MetaWorld MT50 and 94.1% to 98.1% on LIBERO.
-
Yeti: A compact protein structure tokenizer for reconstruction and multi-modal generation
Yeti is a compact tokenizer for protein structures that delivers strong codebook use, token diversity, and reconstruction while enabling from-scratch multimodal generation of plausible sequences and structures with 10x fewer parameters than ESM3.
-
SDFlow: Similarity-Driven Flow Matching for Time Series Generation
SDFlow learns a global transport map via similarity-driven flow matching in VQ latent space, using low-rank manifold decomposition and a categorical posterior to handle discreteness, yielding SOTA long-horizon performance and inference speedups.
-
CapsID: Soft-Routed Variable-Length Semantic IDs for Generative Recommendation
CapsID uses probabilistic capsule routing and confidence-based termination to generate variable-length semantic IDs, improving recall by 9.6% over strong baselines with half the latency of dual-representation systems.
-
FaithfulFaces: Pose-Faithful Facial Identity Preservation for Text-to-Video Generation
FaithfulFaces introduces a pose-faithful identity aligner with a shared dictionary and invariance constraint to maintain facial identity in text-to-video generation under large pose changes and occlusions.
-
Self-Supervised Bootstrapping of Action-Predictive Embodied Reasoning
R&B-EnCoRe uses self-supervised importance-weighted variational inference to distill action-predictive reasoning datasets that improve VLA performance on manipulation, navigation, and driving tasks without external verifiers.
-
Negative Binomial Variational Autoencoders for Overdispersed Latent Modeling
NegBio-VAE introduces negative binomial latents with dispersion to handle overdispersion in discrete VAE models, yielding better reconstruction, generation, and downstream representations than Poisson VAE baselines.
-
IP-Adapter: Text Compatible Image Prompt Adapter for Text-to-Image Diffusion Models
IP-Adapter adds effective image prompting to text-to-image diffusion models using a lightweight decoupled cross-attention adapter that works alongside text prompts and other controls.
-
Rosetta: Composable Native Multimodal Pretraining
Rosetta proposes a composable multimodal pretraining method with MAOP to prevent catastrophic forgetting when expanding modalities beyond standard MoE and MoT approaches.
-
Adversarial Error Correction for Visual Autoregressive Generation
AID-VAR attaches an adversarial discriminator and lightweight guidance injector to frozen VAR backbones to diagnose and correct fidelity gaps across scales, reporting 16% FID gains with 3% added parameters.
-
Atom-level Protein Representation Learning Improves Protein Structure Prediction
TriProRep pretrains on three aligned protein views and improves results on homodimer co-folding and related structure tasks in the new RepSP benchmark.
-
Composable Crystals: Controllable Materials Discovery via Concept Learning
VQ-VAE concept learning enables controllable recombination of crystal motifs to generate structures with reported gains in validity-stability-uniqueness-novelty metrics on MP-20 and Alex-MP-20.
-
A Survey on Vision-Language-Action Models: An Action Tokenization Perspective
The survey frames VLA models as pipelines that generate progressively grounded action tokens and classifies those tokens into eight types to guide future development.