-
The major barrier to efficient parallelization of blocking is the inter process communication of the records. Similarly, much of the potential benefit of parallelization of scoring is lost because of …
-
Skimming the code for `KafkaClient` and `BrokerConnection`, it looks like there's some cleanup possible:
1) I don't see why KafkaClient needs a `check_version()`... why not just completely delegate…
-
Support empty lists in annotations (for NixIO).
See #476
-
--- load weight finish ---
Setting up a new session...
Max_iter = 120000, Batch_size = 20
Model will train on cuda:[0]
--- Focal_loss alpha = 0.25 ,将对背景类进行衰减,请在目标检测任务中使用 ---
--- Multiboxlo…
-
NATTEN ops break graphs because they are not registered as native torch ops.
The correct way to do this since pytorch 2.0 is to register C++ ops through torch library and not torch extensions.
#…
-
BPO | [23808](https://bugs.python.org/issue23808)
--- | :---
Nosy | @pfmoore, @tjguk, @zware, @serhiy-storchaka, @zooba, @izbyshev
PRs | python/cpython#2475
*Note: these values reflect the state of t…
-
```
Building on the type/branch-recording code added in r778, we should
make CALL_FUNCTION and co record which functions they're actually
calling. I see two uses for this data:
a) In the case where…
-
```
Building on the type/branch-recording code added in r778, we should
make CALL_FUNCTION and co record which functions they're actually
calling. I see two uses for this data:
a) In the case where…
-
When generating geojson geometry from shapely the coordinates are tuples and so geojson_rewind fails when it tries to modify the coordinates
```python
from shapely.geometry import Polygon, mappin…
mbway updated
3 years ago
-
WATCHMAN_CHECKS = (
'xyxy.health.xyxy_check'
)
If I have a **single check** like the above, WATCHMAN_CHECKS variable is taken as a string and thus when frozenset is applied in get_checks method…