feat: initialize backend for Hugging Face deployment (by antigravity)

Upload 4 files

feat: initial deploy of clarity hybrid backend with gemini and qwen support

perf: switch to transformers library and native pytorch model for optimized inference

perf: switch to 0.5B model for maximum responsiveness on CPU

perf: switch to 1.5B Q2_K quantization for lowest possible latency on CPU

perf: switch to Qwen 2.5 Coder 1.5B for ultra-fast inference on 2 vCPU hardware

feat: switch to Qwen 2.5 Coder 3B model for faster inference and offline compatibility

Upload 4 files

feat: initial deploy of Clarity backend using Transformers and Qwen 2.5 Coder 3B

Update model context

Upload 4 files