* build: move to uv
* build: add packages declaration
* build: move to python 3.12, as sentencespiece does not work on 3.13
* ci: remove pre-commit check, will be done in another branch.
* ci: fix name checkout
* ci: update lock and dockerfile
* test: remove event_loop, not needed in python 3.12
* test: updated test due to av returning AudioFrame with 4096 samples instead of 1024
* build: prevent using fastapi cli, because there is no way to set default port
I don't want to pass --port 1250 every time, so back on previous
approach. I deactivated auto-reload for production.
* ci: remove main.py
* test: fix quirck with httpx
This feature a new modal endpoint, and a complete new way to build the
summary.
## SummaryBuilder
The summary builder is based on conversational model, where an exchange
between the model and the user is made. This allow more context
inclusion and a better respect of the rules.
It requires an endpoint with OpenAI-like completions endpoint
(/v1/chat/completions)
## vLLM Hermes3
Unlike previous deployment, this one use vLLM, which gives OpenAI-like
completions endpoint out of the box. It could also handle guided JSON
generation, so jsonformer is not needed. But, the model is quite good to
follow JSON schema if asked in the prompt.
## Conversion of long/short into summary builder
The builder is identifying participants, find key subjects, get a
summary for each, then get a quick recap.
The quick recap is used as a short_summary, while the markdown including
the quick recap + key subjects + summaries are used for the
long_summary.
This is why the nextjs component has to be updated, to correctly style
h1 and keep the new line of the markdown.
* serverless: implement banana backend for both audio and LLM
Related to monadical-sas/reflector-gpu-banana project
* serverless: got llm working on banana !
* tests: fixes
* serverless: fix dockerfile to use fastapi server + httpx
