Llama 3.2

The 'Llama 3.2' dataset/benchmark is a synthetically generated dataset designed to evaluate multilingual e-commerce intent recognition using a one-billion-parameter model optimized for performance and resource efficiency.

Papers using Llama 3.2 (4)

Information-Aware KV Cache Compression for Long Reasoning2026BaseRT: Best-in-Class LLM Inference on Apple Silicon via Native Metal2026Performance Trade-offs of Optimizing Small Language Models for E-Commerce2025LLM-NEO: Parameter Efficient Knowledge Distillation for Large Language Models2024