ollama vs ollama_chat provider

[cfytrok]

Why are there 2 types of providers ollama and ollama_chat?
Ollama supports both /api/generate and /api/chat for the same model. Why in litellm I have to create two models to support both /completion and /chat/completion?
For example, I create a model ollama/qwen2.5-coder:0.5b. When calling /chat/completion, the history will be incorrectly rendered in the prompt. Ollama /api/generate method for the chat will be called. I will not see any errors.
If I create ollama_chat/qwen2.5-coder:0.5b, then when calling litellm /completion, ollama /api/chat will be called and the prompt will be rendered as a chat message. Same, no errors.
I think there should be only one ollama provider with supports both /completion and /chat/completion