Pipeline_settings

[AIBappa]

1. When the webapp opens the older instruction to start server using http.server is still present. That needs to be updated.

2. Currently on my windows 11 machine, i just start ollama desktop and connectivity test passes. I have not yet done any addtional settings as shown in pipeline_settings page. So maybe need to check if those are required like setting CORS for ollama. Maybe they can be added in a new seperate troubleshooting section on the same page.

[AIBappa]

Also need to add advanced settings to each local and cloud LLM tab in Pipeline_settings.
This will reflect the following

1. Role
2. Temperature
3. Output format
4. Model-name (Currently only has fixed model names, need custom name)

See also Nvidia NIM format
invoke_url='https://integrate.api.nvidia.com/v1/chat/completions'

payload=$(cat <<'JSON'
{
"model": "deepseek-ai/deepseek-v4-pro",
"messages": [{"role":"user","content":""}],
"temperature": 1,
"top_p": 0.95,
"max_tokens": 16384,
"chat_template_kwargs": {"thinking":false},
"stream": false
}
JSON
)

curl -sS
--request POST
--url "$invoke_url"
--header "Authorization: Bearer $NVIDIA_API_KEY"
--header "Content-Type: application/json"
--data "$payload"

[AIBappa]

Updates have been done in UI Branch but need to be tested.

[AIBappa]

Found certain issues, the red, gray and green lights for connecting up various LLMs is not coded properly. After connection , a green connection shows testing. It should be completed.