Claude 9688e086ca fix: add Metal cache management to prevent memory growth ... - Add ClearCache() wrapping mlx_clear_cache - Clear Metal allocator cache every 8 tokens during generation - Set 16GB cache limit on backend init - Prevents GPU memory from growing unbounded during inference Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>		2026-02-16 05:53:52 +00:00
..
cache	feat: support quantized inference (4-bit) for Gemma 3	2026-02-16 05:53:52 +00:00
model	fix: correct SDPA mask mode and slice logits to last position	2026-02-16 05:53:52 +00:00
sample	fix: remove unused vars in TopP sampler placeholder	2026-02-16 05:53:52 +00:00
tokenizer	fix: handle both string and array merge formats in tokenizer	2026-02-16 05:53:52 +00:00
array.go	fix: correct 20 mlx-c API mismatches for v0.4.1	2026-02-16 05:53:52 +00:00
CMakeLists.txt	chore: target macOS 26.0, fix duplicate -lstdc++ linker warning	2026-02-16 05:53:52 +00:00
compile.go	fix: correct 20 mlx-c API mismatches for v0.4.1	2026-02-16 05:53:52 +00:00
dtype.go	feat: add native MLX backend for Apple Silicon inference (pkg/mlx)	2026-02-16 05:53:52 +00:00
fast.go	fix: correct SDPA mask mode and slice logits to last position	2026-02-16 05:53:52 +00:00
io.go	feat: add native MLX backend for Apple Silicon inference (pkg/mlx)	2026-02-16 05:53:52 +00:00
mlx.go	debug: add shape logging and stderr error handler for inference debugging	2026-02-16 05:53:52 +00:00
mlx_stub.go	feat: add native MLX backend for Apple Silicon inference (pkg/mlx)	2026-02-16 05:53:52 +00:00
nn.go	feat: support quantized inference (4-bit) for Gemma 3	2026-02-16 05:53:52 +00:00
ops.go	fix: use affine quantization mode and infer head_dim from weights	2026-02-16 05:53:52 +00:00
random.go	fix: correct 20 mlx-c API mismatches for v0.4.1	2026-02-16 05:53:52 +00:00
slice.go	fix: correct 20 mlx-c API mismatches for v0.4.1	2026-02-16 05:53:52 +00:00
stream.go	fix: add Metal cache management to prevent memory growth	2026-02-16 05:53:52 +00:00