more readme ideas

README.md

@@ -14,12 +14,15 @@ Check out the configuration reference at https://huggingface.co/docs/hub/spaces-
 This is a test ...
 
 TASKS:
-- rewrite generation from scratch or use the one of mistral space if possible. alternative use https://github.com/abetlen/llama-cpp-python#chat-completion
+- for fast debug: Add a debug mode that enables me to run direct cli commands? -> Never for prod!
+- prod harden docker with proper users etc. OR mention this is only a dev build an intended for messing with, no readonly filesystem etc.
+- rewrite generation from scratch or use the one of mistral space if possible. alternative use https://github.com/abetlen/llama-cpp-python#chat-completion or https://huggingface.co/spaces/deepseek-ai/deepseek-coder-7b-instruct/blob/main/app.py
 - write IN LARGE LETTERS that this is not the original model but a quantified one that is able to run on free CPU Inference 
-- check ho wmuch parallel generation is possible or only one que and set max context etc accordingly. Maybe live-log ram free etc to the interface on "system health" graph if not too resource hungry on its own ... -> Gradio for display??
-- live stream response (see mistral space!!)
-- log memory usage to console? Maybe auto reboot if too slim
-- readd system prompt? maybe checkout how to setup from lm studio - could be a dropdown with an option to set one fix also via env var when only one model available ...
+- test multimodal with llama?
+- can i use swap in docker to maximize usable memory?
+- proper token handling - make it a real chat (if not auto by chatcompletion interface ...)
+- maybe run as webserver locally and gradio only uses the webserver as backend? (better for async but maybe worse to control - just an idea)
+- check ho wmuch parallel generation is possible or only one que and set
 - move model to DL into env-var with proper error handling
 - chore: cleanup ignore, dockerfile etc.
 - update all deps to one up to date version, then PIN them!