LLaVA-1.5-7B

'LLaVA-1.5-7B' is a benchmark used to evaluate the performance of Vision-Language Models (VLMs) in terms of their general understanding and reduction of hallucinations across various tasks.

Papers using LLaVA-1.5-7B (6)

DUET-VLM: Dual stage Unified Efficient Token reduction for VLM Training and Inference2026Self-Correction Inside the Model: Leveraging Layer Attention to Mitigate Hallucinations in Large Vision Language Models2026Cost-Efficient Multimodal LLM Inference via Cross-Tier GPU Heterogeneity2026Token-Level Inference-Time Alignment for Vision-Language Models2025VISOR: Visual Input-based Steering for Output Redirection in Vision-Language Models2025Mitigating Hallucinations via Inter-Layer Consistency Aggregation in Large Vision-Language Models2025