Phi-3-Mini and Llama 3 70B MT-Bench Performance on Long-Context Code Generation

This report synthesises findings from 9 peer-reviewed papers addressing the following research question: How does the score on the MT-bench change for Phi-3-mini versus Llama 3 70B when evaluated on code generation tasks involving long-context reasoning spanning 100K tokens. We introduce phi-3-mini, a 3.8 billion parameter language model trained on 3.3 trillion tokens, whose overall performance, as measured by both academic benchmarks and internal testing, rivals that of models such as Mixtral 8x7B and GPT-3.5 (e.g., phi-3-mini achieves 69\% on MMLU. 15 claims were extracted from source literature; 15 were independently verified against retrieved documents. An automated multi-reviewer quality assessment produced a score of 9.3/10. This report is a machine-generated literature synthesis and does not constitute original research.

Research goal: How does the score on the MT-bench change for Phi-3-mini versus Llama 3 70B when evaluated on code generation tasks involving long-context reasoning spanning 100K tokens?

Autonomous literature synthesis. Automated review score: 9.3/10. Full text and citation available at Assignee Research.