Spaces:
Runtime error
Runtime error
Commit History
-
Fix tool extraction: raw JSON decode for Qwen3 plain output
dexifried71a75b4 -
Force tool use: explicit tool-call rules in all agent prompts
dexifriedb7c551a -
Ensure GPU detection with warmup at init
dexifriedda90885 -
v2.2: FastAPI+Gradio mount, REST /api/agent + /api/health
dexifried018a4d9 -
v2.1: fix generator yield format, single output Chatbot
dexifried083409d -
Add show_error=True for debugging
dexifried25e15a5 -
v2.0: full 5-agent app, Gradio 6.x dict format
dexifried4f0451d -
TEST3: Gradio 6.x dict-based Chatbot format
dexifried2a7dc83 -
TEST2: minimal with @spaces.GPU warmup
dexifried3092a91 -
TEST: minimal echo handler - debugging IndexError
dexifried5f48882 -
Fresh Space: clean build
dexifried6f2acff -
Force rebuild: update docstring
dexifriedfdc43d2 -
Add debug endpoint to test simple handler
dexifried865bc2e
-
Remove backup file
dexifriedd9e961b -
Add try/except with logging to run_agent for debugging IndexError
dexifriedfff84a2 -
Switch to Qwen3-4B (fits in 24GB A10G VRAM)
dexifried473e957 -
Fix: use gr.Blocks.launch(), add REST API via demo.app, add warmup GPU func
dexifriedf1507d9 -
Add REST API endpoint (/api/agent) + client script
dexifried1dede1f -
All-local: Qwen3-30B on H200, DuckDuckGo search, no API keys
dexifried8544029 -
Add top-level @spaces.GPU functions for ZeroGPU detection at startup
dexifried40608b7 -
Fix for Gradio 6.x: move theme/css to launch(), remove deprecated Chatbot type param
dexifried0b90995 -
Fix Gradio 6.0 compatibility: pin to Gradio 5.x, remove deprecated params
dexifried4b95cf8 -
Fix YAML metadata colors
dexifried9496ad0 -
Add HF Spaces YAML metadata
dexifrieda9498ee -
Rewrite: DEX Evolution Outpost — multi-agent orchestration dashboard
dexifried- Command Center with LLM-powered agent routing - Code Agent: stateful Python execution on H200 ZeroGPU - Research Agent: parallel web search + page analysis - Image Agent: Stable Diffusion on ZeroGPU - Web Agent: quick search + URL reading - Dark cyberpunk UI with streaming responses - OpenAI-compatible endpoint support (HF Inference, OpenRouter, etc.)393150e
-
Sovereign Deploy: Live-Streaming H200 Terminal
dexifrieda6ecbb1