Does LiteLLM even have Repetition Penalty?

[BradKML]

Alternate low-temp guidelines for R1 (other than just the recommendation of 0.5-0.7 for temperature):

• From 0.0 for code to 1.0 for data science, nothing other than temperature https://api-docs.deepseek.com/quick_start/parameter_settings
• "Absurdly low" temperature + 1.0 repetition penalty https://www.reddit.com/r/LocalLLaMA/comments/1i7fjqm/comment/m8krz4a/
• 0.0 and DRY to 0.01 https://www.reddit.com/r/LocalLLaMA/comments/1i81ev6/comment/m8shacf/
• 0.1 with 1.2 repetition penalty https://thinhdanggroup.github.io/guide-to-run-deepseek-r1-locally/
• 0.2 with 1.1 repetition penalty https://medium.com/@c.giancaterino/distilled-deepseek-r1-at-work-with-a-naive-rag-614c6e22cf69
• 0.0, with repetition penalty of 1.0 https://www.reddit.com/r/LocalLLaMA/comments/1i7o9xo/deepseek_r1s_open_source_version_differs_from_the/

Some who wants to follow high temp will sometimes recommend mixing in 1.0 repetition penalty (along with the usual mixing in Top P and/or Top K with default values) https://community.appsmith.com/content/guide/building-chat-app-deepseek-r1-and-togetherai-under-5-minutes

Okay so mostly temperature is at 0.0-0.2, with 1.0-1.2 repetition penalty since DRY have low support
Further hints on how for math and coding, keeping it lower than general reasoning recommendations (less than 0.6) are still important https://www.reddit.com/r/LocalLLaMA/comments/1ias719/comment/m9cknjb/
But at the table there is no mention of repetition penalty only presence_penalty and frequency_penalty https://docs.litellm.ai/docs/completion/input
P.S. Does anyone want to support DRY and other methods of preventing repetition? It's like asking for Min P instead of Top P