SorcererLM 8x22B

SorcererLM 8x22B, provided by raifle and created on November 8, 2024, is an advanced RP and storytelling model fine-tuned on WizardLM-2 8x22B, featuring a 16000 context length. The model demonstrates competitive response times, ranking in the 54th percentile for speed across five benchmarks. However, it tends to be positioned at premium pricing levels, falling into the 20th percentile for cost. A significant strength is its strong reliability, achieving an 83% success rate across benchmarks, indicating consistent delivery of usable responses. In terms of benchmark performance, SorcererLM shows varied results. It struggles with hallucination detection and general knowledge, scoring 76.7% and 37.0% accuracy respectively, placing it in the lower percentiles for these categories. Similarly, its ethics and reasoning capabilities are moderate, with 75.0% and 36.6% accuracy. A notable strength lies in its Email Classification, where it achieved 96.0% accuracy, suggesting proficiency in contextual understanding and categorization. Overall, while excelling in specific classification tasks and maintaining high reliability, the model exhibits weaknesses in factual recall, ethical reasoning, and hallucination avoidance.