Complexity Scorer
A request-complexity scoring component based on a fine-tuned small model that outputs a difficulty score between 0 and 1 for each prompt, used to route requests to models of differing capability.
Timeline (last 90 days)
The developer fine-tuned Qwen3-8B into a model called Complexity Scorer that assigns prompts a complexity score from 0.00 to 1.00 for LLM routing.
On a fixed 50-sample evaluation set, the Qwen3-8B model fine-tuned on 50 examples achieved 86% accuracy, higher than the 76% of the version fine-tuned on 2,500 examples.
All Facts (2)
The developer fine-tuned Qwen3-8B into a model called Complexity Scorer that assigns prompts a complexity score from 0.00 to 1.00 for LLM routing.
50%UnverifiedOn a fixed 50-sample evaluation set, the Qwen3-8B model fine-tuned on 50 examples achieved 86% accuracy, higher than the 76% of the version fine-tuned on 2,500 examples.
50%