Tag: Multimodal

8 articles

AI Research 42 - Multimodal Large Model Quantization: From FP32 to INT4, the Final Summary

Survey outline for multimodal large model quantization schemes: from FP32 to INT4. Core goal is model capability retention, compression efficiency 50-75%...

AI Research 41 - Multimodal Large Model Quantization: Qwen2.5-VL Architecture, Capability Evaluation and Use Cases

Qwen2.5-VL is the new generation multimodal large model launched by Alibaba, significantly leading in visual understanding, video analysis, and cross-modal reasoning.

AI Research 40 - Multimodal Large Model Quantization: Landscape Reshaping Through Five Open-Source Models

Multimodal large models are developing rapidly, with representative models like BLIP-2, MiniGPT-4, Flamingo, LLaVA, and Qwen2.5-VL emerging.

AI Research 39 - Multimodal Large Model Quantization: How Fine-Tuning and Quantization Maximize Performance

In multimodal large model optimization, the order choice of fine-tuning and quantization directly affects the final model's performance and efficiency.

AI Research 38 - Multimodal Large Model Quantization: Evaluation Strategies for Mainstream Vision-Language Tasks

To systematically evaluate the impact of model quantization on performance, need to combine multiple vision-language datasets and metrics.

AI Research 37 - Multimodal Large Model Quantization: Impact on Vision, Language and Multimodal Tasks

Model quantization compresses FP32 weights into low-precision representations, significantly reducing inference resource consumption.

AI Research 36 - Comprehensive Analysis of Multimodal Large Model Quantization

This comprehensive overview systematically introduces mainstream quantization techniques in multimodal models.

AI Research 13 - LLM and Agent Research: The Rise and Development of LLM Agents

2024 is called the 'Year of Agents'. LLM trends show parallel development of 'bigger and stronger' and 'smaller and more specialized'.