moose Claude Opus 4.7 (1M context) commited on
Commit ·
64a4e11
1
Parent(s): fcd1c32
Lower AOTI compile duration 1500 → 1200 (cap probe)
Browse filesZeroGPU is rejecting duration=1500 with "larger than the maximum allowed"
despite the AOTI blog still documenting 1500s as the startup max. Bisecting
to find the current per-call cap.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- optimization.py +1 -1
optimization.py
CHANGED
|
@@ -49,7 +49,7 @@ INDUCTOR_CONFIGS = {
|
|
| 49 |
|
| 50 |
def optimize_pipeline_(pipeline: Callable[P, Any], *args: P.args, **kwargs: P.kwargs):
|
| 51 |
|
| 52 |
-
@spaces.GPU(duration=
|
| 53 |
def compile_transformer():
|
| 54 |
|
| 55 |
# Run warmup pass to capture transformer inputs (needs all components on GPU)
|
|
|
|
| 49 |
|
| 50 |
def optimize_pipeline_(pipeline: Callable[P, Any], *args: P.args, **kwargs: P.kwargs):
|
| 51 |
|
| 52 |
+
@spaces.GPU(duration=1200)
|
| 53 |
def compile_transformer():
|
| 54 |
|
| 55 |
# Run warmup pass to capture transformer inputs (needs all components on GPU)
|