LiteLLM Support (Also DeepSeek in the UI and Llamafile in the UI) (#911)

* start litellm support, not yet working

* Remove all the completions route code

* update walkthrough docs

* Working and added Deepseek support!

* Deepseek comparison

* Explanation of pre-reqs

* Apply suggestions from code review

Co-authored-by: Hareesh  <hareeshbahuleyan@gmail.com>
Signed-off-by: Nathan Brake <33383515+njbrake@users.noreply.github.com>

* merged

* First attempt at a parametrized JobCreate

* Replace templates with pydantic models

* Adapt SDK and SDK tests

* Fix sdk unit tests

* Fix notebook tests

* Fix tests

* Fix job definition in workflows

* Fix job unit test

* Start a default workflow for experiments

* Rebase to main

* delete the evaluator

* Rebase to main

* First attempt at a parametrized JobCreate

* Replace templates with pydantic models

* Adapt SDK and SDK tests

* Fix sdk unit tests

* Fix notebook tests

* Fix tests

* Fix job definition in workflows

* Fix job unit test

* Start a default workflow for experiments

* Rebase to main

* Align with routes in main

* Move to experiments new endpoint

* Streamline new experiments api

* Add dataset/samples to experiment, test background tasks

* Factor out experiment formatting

* Support Llamafile

* cleanup for PR

* Clean up for PR

* Apply suggestions from code review

Co-authored-by: Dimitris Poulopoulos <dimitris@mozilla.ai>
Signed-off-by: Nathan Brake <33383515+njbrake@users.noreply.github.com>

* Updates from PR, remove dup models.json file

* fix enable_tqdm schema

* Update naming

* Fix a bug with status

* Exclude unset and none

* job unit test and jupy notebook patch

* job unit test and jupy notebook patch

* final pass through the code

* clean notebook output

---------

Signed-off-by: Nathan Brake <33383515+njbrake@users.noreply.github.com>
Co-authored-by: Hareesh <hareeshbahuleyan@gmail.com>
Co-authored-by: Javier Torres <javier@mozilla.ai>
Co-authored-by: Dimitris Poulopoulos <dimitris@mozilla.ai>

README.md

@@ -59,8 +59,8 @@ services networked together to make up all the components of the Lumigator appli
 > uses SQLite for this purpose.
 
 > [!NOTE]
-If you want to evaluate against LLM APIs like OpenAI and Mistral, you need to set the appropriate
-environment variables: `OPENAI_API_KEY` or `MISTRAL_API_KEY`. Refer to the
+If you want to evaluate against LLM APIs like OpenAI/Mistral/Deepseek, you need to set the appropriate
+environment variables: `OPENAI_API_KEY` or `MISTRAL_API_KEY` or `DEEPSEEK_API_KEY`. Refer to the
 [troubleshooting section](https://mozilla-ai.github.io/lumigator/get-started/troubleshooting.html#tokens-api-keys-not-set)
 in our documentation for more details.
 
@@ -87,10 +87,11 @@ To start Lumigator locally, follow these steps:
     **Important: Continue the next steps in this same terminal.**
 
 
-1. If you intend to use Mistral API or OpenAI API, use that same terminal and run:
+1. If you intend to use Mistral API or OpenAI API or Deepseek API, use that same terminal and run:
     ```bash
     export MISTRAL_API_KEY=your_mistral_api_key
     export OPENAI_API_KEY=your_openai_api_key
+    export DEEPSEEK_API_KEY=your_deepseek_api_key
     ```
     **Important: Continue the next steps in this same terminal.**