leafspark/Llama-3.1-8B-MultiReflection-Instruct

leafspark/Llama-3.1-8B-MultiReflection-Instruct is an 8 billion parameter Llama-3.1-based instruction-tuned model developed by leafspark, inspired by OpenAI's o1 reasoning model. It is fine-tuned for advanced agentic reasoning, generating verbose, multi-step thought processes and reflections in XML format. The model excels at tasks requiring detailed, coherent reasoning and is optimized for a 32768-token context length.