Optimize Qwen with streaming and auto device map

Restore torch_dtype='auto' for optimization

Revert optimization: Back to float32

Optimize for speed: Auto precision & low throttle

Add optimized streaming endpoint to app.py

Add streaming support to app.py

REVERT to Qwen2.5-Coder-1.5B (Restore Stable Deployment)

Add setuptools and wheel to fix build error

RESTORE Dolphin Uncensored Model (User Requested)

Restore stable Qwen 1.5B model

Switch to Dolphin Llama 3.2 1B (Uncensored)

Switch to Transformers + Qwen2.5-Coder-1.5B for instant build

Switch to pre-built wheels for faster build

Add API documentation

Revert Space Title to personal-coder-ai

Update Space Title to Loomyloo AI (Qwen-14B)

Add Dockerfile to fix build errors

Switch to Qwen2.5-Coder-14B-Instruct-Uncensored GGUF

Upgrade to Qwen2.5-1.5B-Instruct for better logic

Upgrade to Qwen2.5-0.5B-Instruct for smarter memory

Fix chat template generation prompt

Add conversation memory and reset button

Switch to SmolLM2-135M-Instruct for smarter and faster responses

Fix weird AI responses with prompt template and repetition penalty

Upgrade to flan-t5-large and improve generation parameters

Add sentencepiece and accelerate for Flan-T5

Update Dockerfile to use uvicorn directly and add startup logs

Force rebuild with v2 UI

Add Chat UI and enable CORS

Deploy Flan-T5 Docker with FastAPI Standard

Fix requirements for FastAPI

Deploy Google Flan T5 FastAPI Docker app

Upload 4 files

Upload 4 files

Upload 2 files

Upload 4 files

Upload Dockerfile

Upload Dockerfile

Upload 2 files

Upload 2 files

Upload 4 files

initial commit