Prompt Engineered GPT-4 Beats Gemini on all of Google's text benchmarks
by sp332 on 12/12/2023, 4:06:28 PM
"We note that Medprompt+ relies on accessing confidence scores (logprobs) from GPT-4. These are not publicly available via the current API but will be enabled for all in the near future."
Had that been announced yet?
by alecco on 12/12/2023, 9:33:39 PM
It would be interesting to see how this can be used for open models.
by byyoung3 on 12/12/2023, 11:06:24 PM
engineered? Over overfit?
"We note that Medprompt+ relies on accessing confidence scores (logprobs) from GPT-4. These are not publicly available via the current API but will be enabled for all in the near future."
Had that been announced yet?