qwen 3.6 qwen3.6 alibaba open source llm large language model github download repository hugging face qwen3.6-27b qwen3.6-35b-a3b mixture of experts moe gated delta networks hybrid attention architecture mtp speculative decoding 262k context window 1m context yarn rope scaling agentic coding frontend workflows

This project showcases fine-tuning of the Qwen2.5-3B model using the Chain-of-Thought (CoT) prompting technique to enhance mathematical reasoning capabilities. With efficient parameter-efficient fine-tuning (PEFT) via LoRA and the high-speed Unsloth framework, the model is optimized for clarity, depth, and accuracy in problem-solving.

О текстиле

Access the Qwen 3.6 LLM via a cross-platform desktop application for chat and coding tasks.