commitcopilot/infer-001

chore(config): update llama server arguments

chore(config): switch default model to Gemma

perf(config): increase model batch size

docs(readme): remove default Spaces reference

chore(config): update default model artifact

feat(llama): add parallel server configuration

feat(generation): disable model reasoning by default

feat(config): enable prompt caching by default

chore(config): tune model validation and llama defaults

refactor(api): structure app config and lifecycle

feat(api): support Qwen tool calls

fix(api): use configurable readiness wait settings

chore(config): switch default model to Gemma E2B

chore(config): tune llama runtime defaults

feat(config): increase llama context window

fix(api): return requested model in generate response

fix(api): allow non-integer usage values

fix(model): validate downloaded GGUF files

feat(inference): run generation through llama-server

build(docker): add musl runtime dependencies

build(docker): use prebuilt llama.cpp wheels

build(docker): add pkg-config dependency

feat(inference): add cloud inference worker

initial commit