abacusai/Smaug-Qwen2-72B-Instruct

The Smaug-Qwen2-72B-Instruct is a 72.7 billion parameter instruction-tuned causal language model developed by abacusai, fine-tuned from Qwen2-72B-Instruct. This model is optimized for complex reasoning and problem-solving tasks, demonstrating improved performance on benchmarks like Big-Bench Hard (BBH), LiveCodeBench, and Arena-Hard compared to its base model. With a substantial 131,072 token context length, it is well-suited for applications requiring deep contextual understanding and advanced analytical capabilities.