A new survey paper published on arXiv details methods for accelerating inference in Video Large Language Models (VLMs). The paper categorizes techniques based on their intervention points within the inference pipeline, including input processing, visual encoding, projector, prefill, and decoding stages. It also discusses Post-Training Quantization (PTQ) as a cross-stage method and analyzes the trade-offs between computational savings and the preservation of spatiotemporal evidence. AI
IMPACT Provides a structured overview of techniques to reduce computational costs for video understanding models.
RANK_REASON Academic paper published on arXiv detailing methods for accelerating VLMs. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- arXiv
- CatalyzeX Code Finder for Papers
- computer science
- Computer vision and pattern recognition
- Connected Papers
- CORE Recommender
- DagsHub
- Gotit.pub
- Hugging Face
- Influence Flower
- Litmaps
- Post Training Quantization Preprocessing Method of Convolutional Neural Network via Outlier Removal
- ScienceCast
- scite Smart Citations
- Video Large Language Models
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
