v0.0.171

refactor: adjust context size

Signed-off-by: thxCode <[email protected]>

v0.0.170

1. Rebase upstream.
2. Fix tool calling with --jinja in Qwen3 Coder.

v0.0.169

1. Rebase upstream.
2. Fix MUSA release suffix.

v0.0.168

1. Rebase upstream.

v0.0.167

1. Rebase upstream;
2. Bump MUSA to rc4.2.0, cc @yeahdongcn .

v0.0.166

1. Rebase upstream.

v0.0.165

1. Rebase upstream;
2. Support Kimi-K2.

v0.0.164

1. Rebase upstream;
2. Fix zero offloading VRAM occupied in DL packages.

v0.0.163

1. Fix failed while chatting service deployed inside a Docker container.

v0.0.162

1. Simplify samplers;
2. Refactor managing embeddings KV cache;
3. Fix invalid capacity in Darwin RPC server.