Qwen2.5-HA-0.5B-Instruct

Introduction

Qwen2.5-HA-0.5B-Instruct is a smart home model fine-tuned based on Qwen2.5-0.5B-Instruct, with approximately 500 million parameters. The main features of this model include:

• Model Type: Causal Language Model
• Training Stages: Pre-training and post-training
• Architecture: Transformer, using RoPE, SwiGLU, RMSNorm, Attention QKV bias, and tied word embeddings
• Number of Parameters: 490 million (360 million non-embedding parameters)
• Number of Layers: 24 layers
• Number of Attention Heads (GQA): 14 query heads, 2 key-value heads
• Context Length: Supports a full 32,768 tokens, with a maximum generation of 8,192 tokens

This model shows significant improvements in instruction understanding, long-text generation, and structured data comprehension. It supports multilingual capabilities across 29 languages, including English, Chinese, and French. The model has been fine-tuned on smart home datasets and can produce structured outputs by setting appropriate system prompts.

Available NPU Models

Home Assistant Fine-tuned Models

qwen2.5-ha-0.5b-ctx-ax630c

• Compared to the base model, it provides a longer context and stable structured output in Home Assistant–specific JSON format
• Supports a context window length of 1024
• Maximum output of 1280 tokens
• Supported platforms: LLM630 Computing Kit, Module LLM, and Module LLM Kit
• TTFT: 533.19 ms
• Average generation speed: 9.76 token/s

Installation

apt install llm-model-qwen2.5-0.5b-int4-ax630c

• Download llm-model-qwen2.5-0.5b-int4-ax630c

qwen2.5-ha-0.5b-ctx-ax650

• Compared to the base model, it provides a longer context and stable structured output in Home Assistant–specific JSON format
• Supports a context window length of 1024
• Maximum output of 1280 tokens
• Supported platform: AI Pyramid
• TTFT: 143.02 ms
• Average generation speed: 25.5 token/s

Installation

apt install llm-model-qwen2.5-ha-0.5b-ctx-ax650

• Download llm-model-qwen2.5-ha-0.5b-ctx-ax650