Overview

meshllm/llama-3.2-1b-instruct-parity-bf16-mlx is a 1 billion parameter instruction-tuned language model based on the Llama 3.2 architecture. This specific version is an MLX artifact, converted from the meta-llama/Llama-3.2-1B-Instruct checkpoint into the bf16 (bfloat16) format. Its primary purpose is to serve as a parity testing model within the mesh-llm ecosystem.

Key Characteristics

• Architecture: Llama 3.2
• Parameter Count: 1 billion (1B)
• Format: MLX bf16 (bfloat16)
• Origin: Converted from meta-llama/Llama-3.2-1B-Instruct
• Context Length: 32768 tokens

Intended Use and Validation

This model is explicitly designed for parity testing against its intended pair, meshllm/llama-3.2-1b-instruct-parity-f16-gguf. It has been validated using the mesh-llm exact smoke suite, confirming expected behavior including shared strict lowercase drift on specific prompts (blue/green/red) and clean agreement on other exact prompts. Developers can use this model to ensure consistent output and behavior across different quantization and framework conversions within the mesh-llm framework, particularly for MLX-based applications.