[Feature Request] Support chat format in the llm generate APi

[snowmanwwg]

The LLM generate API (https://github.com/NVIDIA/NeMo/blob/93a77b1d56df80ca7f772ad37563ab1ec39a21b6/nemo/collections/llm/api.py#L1037) fails with chat-format data.

[athitten]

This will be supported in Megatron Bridge's generate method. Assigning this to @pthombre who is working on adding generate functionality to Megatron Bridge.

[athitten]

@pthombre to check with MBridge team on this if they can support chat template in the generate API. MBridge does not have generate method yet.

[athitten]

Ananth is adding generate functionality to MBridge. @pthombre to get more context on if chat template can also be supported.