HF RL Explorer

I want the Python client objects to support streamed Ollama responses when I opt in with stream=True .

I want the Python client objects to support streamed Ollama responses when I opt in with stream=True .: a task in MiMo-V2.6-RL-harbor-code: MiMo-V2.6-RL Code (Harbor) (Harbor dataset). For example, if /api/generate streams newline-delimited JSON objects {"model":"dummy","response":"Because "} …

The task

For example, if `/api/generate` streams newline-delimited JSON objects `{"model":"dummy","response":"Because "}`, `{"model":"dummy","response":"it "}`, and `{"model":"dummy","response":"is."}`, then `for part in Client(host).generate('dummy', 'Why is the sky blue?', stream=True)` should yield three `GenerateResponse`…

Part of FineEnvs/MiMo-V2.6-RL-harbor-code.