Modelauto-detected

meta-llama/Llama-3.2-11B-Vision

Also known as: Llama-3.2-11B-Vision

1stories this week
1last 30 days
1all time

In the model registry

ModelParamsContextReleased
meta-llama/Llama-3.2-11B-Vision10.6B–Sep 18, 2024

Timeline

  1. Oct 15, 2026 · Research paper · 1 source
    MYOPIA: Understanding the Limits of Tree-Based Reasoning in Vision-Language Models
    We introduce MYOPIA, a benchmark comprising four LLM-generated, human-refined decision trees for GTSRB, MNIST, CIFAR-10, and MSCOCO, together with matched prompts, diagnostic protocols, and an evaluation harness.

Often appears with