sway Public

Notifications Fork 0 Star 0
trunk
Branches trunk
1 Branches 0 Tags
Go to file T
Code
Matthew Forrester Wolffe Merge pull request #32 from tenseleyFlow/sprint/36-sway-serve
f04e68b 476 Commits
__init__.py backends: build() dispatches kind='api' to ApiScoringBackend
_instrumentation.py backends/_instrumentation: BackendStats batching counters + cached_batch entry
_mlx_convert.py backends/_mlx_convert: PEFT→MLX-LM LoRA converter (F01)
api.py backends: dummy/api/mlx loop over self.next_token_dist so subclass overrides fire
dummy.py backends: dummy/api/mlx loop over self.next_token_dist so subclass overrides fire
hf.py ruff auto-format after S23 edits (line-length-driven, no logic)
mlx.py backends/mlx: cast logits to fp32 before numpy handoff (bf16 buffer fix)
two_model.py backends/two_model: compose safe_for_concurrent_views from inner backends (F06)