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

fix: install libgomp1 runtime dependency for OpenMP support in final Docker stage

fix: resolve libc.musl error by compiling llama-cpp-python from source in builder stage

fix: optimize backend Dockerfile with multi-stage build and pre-built wheels

Upload 4 files

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

Upload 4 files