YouLearn-Qwen3-0.6B-V11-GGUF

YouLearn Qwen3 0.6B V11 - GGUF

Fine-tuned Qwen3 0.6B for the YouLearn offline study assistant Android app. Quantized to Q4_K_M for mobile deployment.

Model Details

• Base Model: Qwen/Qwen3-0.6B
• Fine-Tuning: YouLearn V11 instructional dataset, 350 training steps
• Quantization: Q4_K_M (4-bit medium)
• Primary Use Case: Ultra-low-RAM (~4GB) Android devices — fallback study assistant

Purpose

This is the smallest model in the YouLearn lineup, designed for devices with limited RAM (~4GB). It serves as a reliable fallback when larger models cannot be loaded.

Capabilities

• Basic study Q&A
• Simple document-grounded answers
• Structured output (key points, flashcards)
• Fallback "Not enough source text" behavior for weak evidence

Usage (llama.cpp)

./llama-cli -m YouLearn-Qwen3-0.6B-V11-FT350-Q4_K_M.gguf \
  -p "<|im_start|>system\nYouLearn is a private offline study tutor.\n<|im_end|>\n<|im_start|>user\nExplain photosynthesis.<|im_end|>\n<|im_start|>assistant\n" \
  -n 256 \
  --top-p 0.85 --temp 0.35 --min-p 0.05

Limitations

• 0.6B parameter model — limited reasoning depth compared to 1B+ models
• May produce shorter/simpler answers than the primary MiniCPM5 1B model
• Designed as a fallback, not a primary study assistant