ruggsea/Llama3-philosophy-demo

Reduce GPU reservation from 120s to 60s per call

Add /chat_with_history API endpoint for multi-turn conversations

Fix prefix stripping: handle at chunk level before accumulation

Strip 'assistant' prefix artifact from model responses

fix: revert to working minimal version (no additional_inputs)

feat: restore additional inputs (system prompt, sliders) with working API

fix: simplify to minimal ChatInterface to debug zero endpoints issue

fix: add explicit type='messages' for Gradio 5 compatibility

fix: match official HF ZeroGPU pattern exactly

fix: use .to('cuda') instead of device_map='auto' for ZeroGPU compatibility

fix: migrate to ZeroGPU (free H200), remove broken bnb 4-bit quantization

req

push

redesign

Fixing the chat history

roba

fixing

fix

fixing

fix

fixing

fixed structure

style

update requirements

update

Fixing css

fix scrolling

reverted

fix

requirements

changed model, added system prompt

changing ui

Updated to use ruggsea/Llama3.1-8B-SEP-Chat with multi-turn support

Update app.py

Update README.md

Update app.py

Update app.py

Update app.py

Update app.py

Adapting space to philosophy model

Update

Update

Upgrade Gradio to version 4.26.0 (#52)

Update

Update

Update

Update

Update

Update

Update