driaforall/mem-agent

The driaforall/mem-agent is a 4 billion parameter language model based on Qwen3-4B-Thinking-2507, specifically trained using GSPO for agentic interactions with an Obsidian-like memory system. It excels at retrieval, updating, and clarification tasks within this memory framework, utilizing a suite of file and directory manipulation tools. This model is optimized for applications requiring sophisticated memory management and agentic reasoning, demonstrating strong performance in memory-centric benchmarks.