-
So we currently have `number` to support floating point values. These are implemented as `float` which is a 32-bit floating-point implementation. The issue we have is of rounding, and of how to repres…
-
Hi, thanks for your work. I'm trying your fa2-cm but it raises error because of the following assertion:
```python
assert do.is_contiguous()
assert q.stride() == k.stride() == v.stride() == o.strid…
-
### Search before asking
- [X] I have searched the Ultralytics YOLO [issues](https://github.com/ultralytics/ultralytics/issues) and [discussions](https://github.com/ultralytics/ultralytics/discussion…
-
(cc @lezcano)
I did some experimenting regarding 3xTF32 precision in Triton. I used matmul tutorial code, changed to use 3xTF32 calculation, and compared its precision to the precision of 3xTF32 im…
-
Omega_prec determins the number of siginificant figures for the output. The default 1e-3 has proven to give large discontinuities in UQ scales and could potentially cause issues in DEEPlasma too. I pr…
-
Some of the derived quantities are producing double precision values rather than single precision, which causes errors in some cases. Which features are doubles depends on which systematics are used. …
-
Problem description:
Loading up a video with NotchLC Codec and alpha channel displays unexpected results.
The video itself is simply a moving color shape with an alpha gradient.
Playing with th…
-
When converting an `Instant` to `NSDate` and back to `Instant` the value is rounded to milliseconds. However, when testing I see microsecond precision after calling `toNSDate()` and even when using `N…
-
When changing the precision configuration, whether the stopwatch was running is maintained (the `timer` property), but the `watch` is reset. That's inconsistent at least, but one might argue what's th…
-
### System Info
- `transformers` version: 4.44.1
- Platform: Linux-5.15.0-88-generic-x86_64-with-glibc2.35
- Python version: 3.10.12
- Huggingface_hub version: 0.24.6
- Safetensors version: 0.4.4…