chore: (FTUE ) updated display pipeline for builder and interface like objects

[nabinchha]

Part 1 of FTUE

• Make info pipeline extensible across config and interface objects.

from data_designer.essentials import (
    DataDesigner,
    DataDesignerConfigBuilder,
    SamplerType,
    InfoType,
)

data_designer = DataDesigner()

# Display default model providers. At this point it will print out NVIDIA provider info
# If it detects `NVIDIA_API_KEY` is not present, a warning log message is printed out
data_designer.info.display(InfoType.MODEL_PROVIDERS)

config_builder = DataDesignerConfigBuilder(model_configs=data_designer.get_default_model_configs())

# Display all samplers 
config_builder.info.display(InfoType.SAMPLERS,)

# Display a specific sampler
config_builder.info.display(InfoType.SAMPLERS, sampler_type=SamplerType.BERNOULLI)

# Display model configs
config_builder.info.display(InfoType.MODEL_CONFIGS)

[github-actions]

All contributors have signed the DCO ✍️ ✅
_{Posted by the DCO Assistant Lite bot.}

[nabinchha]

These will be updated in the next PR to pull from yml config files written by the cli.

[nabinchha]

I have read the DCO document and I hereby sign the DCO.

[nabinchha]

recheck

[nabinchha]

recheck

[johnnygreco]

@kirit93 – please chime in with thoughts on default models.

[nabinchha]

We all chatted offline to support multiple provider (nvidia, opena) to start with along with their model configs. It' all handled in this commit:

Behaviors:

1. This displays the default model providers (nvidia and and openai). If the required API key cannot be found, it will show it's missing.

data_designer = DataDesigner()
data_designer.info.display(InfoType.MODEL_PROVIDERS)

2. This displays the available default model configs based on which API keys were detected. In this example, only openai model configs were shown because nvidia api key could not be found. There's a warning that gets printed out as well.

config_builder = DataDesignerConfigBuilder(model_configs=data_designer.get_default_model_configs())
config_builder.info.display(InfoType.MODEL_CONFIGS)

3. If both nvidia and openai models cannot be found, we'll see warnings for each of the keys not being found + additional warnings.

[nabinchha]

@johnnygreco if you are cool with the above, let's get this merged in. We can further tweak the model names, etc in follow up PRs.

[johnnygreco]

Yeah, sounds good. We can tweak in follow ups. One thought I had was to have get_default_model_configs take an argument ("nvidia" or "openai"), which we can default to "nvidia". That way we can only raise a warning for the appropriate API key when needed.

[nabinchha]

Yea sounds good I'll get that in the next PR.

[nabinchha]

@mikeknep just want to make sure you are are of these changes for the nmp dd client.