Add 'parameters:' key in transform_input function for llama2

[johnphilips3]

I deployed this sample with a meta-textgeneration-llama-2-7b-f endpoint generated from Jumpstart and was getting this error from endpoint due to incorrectly formatted input:

ValueError: Error raised by inference endpoint: An error occurred (ModelError) when calling the InvokeEndpoint operation: Received client error (424) from primary with message "{
  "code":424,
  "message":"prediction failure",
  "error":"Input payload must contain a 'inputs' key and optionally a 'parameters' key containing a dictionary of parameters."

This change adds a 'parameters:' key to the model_kwargs value and corrects the error.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[MithilShah]

Llama2 7B works differently from 70B. will have to add 7B as a new feature.