MiVE repurposes VLMs as multiscale feature extractors integrated into a unified self-attention Diffusion Transformer for reference-guided video editing, claiming top human preference scores over prior methods.
Title resolution pending
7 Pith papers cite this work. Polarity classification is still indexing.
7
Pith papers citing it
representative citing papers
Instructions trigger a production-centered mechanism in language models, with task-specific information stable in input tokens but varying strongly in output tokens and correlating with behavior.
SAMoRA is a parameter-efficient fine-tuning framework that uses semantic-aware routing and task-adaptive scaling within a Mixture of LoRA Experts to improve multi-task performance and generalization over prior methods.
Uptraining multi-head transformer checkpoints to grouped-query attention models achieves near multi-head quality at multi-query inference speeds using 5% additional compute.
CodeT5 adds identifier-aware pre-training and bimodal dual generation to a T5-style encoder-decoder, yielding better results on defect detection, clone detection, and code-to-text, text-to-code, and code-to-code tasks than prior encoder-only or decoder-only models.
Quantized LLMs diverge from their base models at the decision level even when accuracy is preserved, with query and key attention projections showing the greatest structural distortion under low-bit compression.
Hidden layer distillation yields systematic perplexity gains over logit KD in LLM pre-training but does not consistently improve downstream performance.
citing papers explorer
-
MiVE: Multiscale Vision-language features for reference-guided video Editing
MiVE repurposes VLMs as multiscale feature extractors integrated into a unified self-attention Diffusion Transformer for reference-guided video editing, claiming top human preference scores over prior methods.
-
Instructions Shape Production of Language, not Processing
Instructions trigger a production-centered mechanism in language models, with task-specific information stable in input tokens but varying strongly in output tokens and correlating with behavior.
-
SAMoRA: Semantic-Aware Mixture of LoRA Experts for Task-Adaptive Learning
SAMoRA is a parameter-efficient fine-tuning framework that uses semantic-aware routing and task-adaptive scaling within a Mixture of LoRA Experts to improve multi-task performance and generalization over prior methods.
-
GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints
Uptraining multi-head transformer checkpoints to grouped-query attention models achieves near multi-head quality at multi-query inference speeds using 5% additional compute.
-
CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation
CodeT5 adds identifier-aware pre-training and bimodal dual generation to a T5-style encoder-decoder, yielding better results on defect detection, clone detection, and code-to-text, text-to-code, and code-to-code tasks than prior encoder-only or decoder-only models.
-
The Illusion of Equivalency: Statistical Characterization of Quantization Effects in LLMs
Quantized LLMs diverge from their base models at the decision level even when accuracy is preserved, with query and key attention projections showing the greatest structural distortion under low-bit compression.
-
A Study on Hidden Layer Distillation for Large Language Model Pre-Training
Hidden layer distillation yields systematic perplexity gains over logit KD in LLM pre-training but does not consistently improve downstream performance.