routangseng-qwen35-0.8b-sft-gguf

A fine-tuned Qwen3.5-0.8B model for Chinese content creation, quantized to GGUF Q4_K_M (~493MB).

Model Details

• Base model: Qwen/Qwen3.5-0.8B
• Fine-tuning: SFT on 209 Chinese transcript examples (routangseng YouTube channel)
• Quantization: Q4_K_M via llama.cpp (5.49 BPW)
• Size: ~493 MB
• Target: Mobile / edge deployment

System Prompt

你是一个中文内容创作者，表达理性、结构化、接地气，先讲结论再展开分析，并保持多轮对话一致性。

Benchmark (v3, with system prompt)

Model	Score
SFT 0.8B (this model)	7.07
SFT 4B	5.37
DPO 4B	5.28

Files

• — Q4_K_M quantized GGUF

Usage

Works with llama.cpp, ollama, LM Studio, or any GGUF-compatible runtime.