T3Q-LLM1-CV / README.md
JungZoona's picture
Update README.md
c53b0fd verified
|
raw
history blame
797 Bytes

Feature Porting

  • informative_model_name: "upstage/SOLAR-10.7B-Instruct-v1.0"
  • base_model_name: "upstage/SOLAR-10.7B-v1.0"
  • target_model_name: "yanolja/EEVE-Korean-Instruct-10.8B-v1.0"
Task Version Metric Value Stderr
kobest_boolq 0 acc 0.9466 ± 0.0060
kobest_boolq 0 macro_f1 0.9466 ± 0.0060
kobest_copa 0 acc 0.7730 ± 0.0133
kobest_copa 0 macro_f1 0.7727 ± 0.0132
kobest_hellaswag 0 acc 0.5120 ± 0.0224
kobest_hellaswag 0 acc_norm 0.5460 ± 0.0223
kobest_hellaswag 0 macro_f1 0.5099 ± 0.0224
kobest_sentineg 0 acc 0.8086 ± 0.0198
kobest_sentineg 0 macro_f1 0.8071 ± 0.0200