electroglyph/Qwen3-4B-Instruct-2507-uncensored-unslop-v2

The electroglyph/Qwen3-4B-Instruct-2507-uncensored-unslop-v2 is a 4 billion parameter instruction-tuned causal language model based on the Qwen3 architecture. Developed by electroglyph, this model is a GRPO finetune specifically designed to mitigate 'slop' or verbose, repetitive output often found in uncensored models. It offers a distinct style compared to standard Qwen3 4B models, making it suitable for applications requiring concise and direct responses.