-
Platforms: rocm
This test was disabled because it is failing in CI. See [recent examples](https://hud.pytorch.org/flakytest?name=test_flash_attention_vs_math_ref_grads_batch_size_1_seq_len_q_1024_seq…
-
### Issue type
Bug
### Have you reproduced the bug with TensorFlow Nightly?
No
### Source
binary
### TensorFlow version
2.14.0
### Custom code
Yes
### OS platform and distribution
Linux
##…
-
Hi, I want to report a bug and corresponding fix to your team.
I believe this fix could help others who use WGAN-gp algorithm to train GANs.
## Bug and Reason
It appears in the "code example" …
-
In the `examples/application/federated-learning/` , I study the `run.py`. In the function `federated_training` as following:
```
@tf.function
def federated_training(model_owner, data_owners):
…
-
Take args.iter_size==2 for example, I think the clipped and accumulated grads of your codes are clip(clip(grads1)+grads2), not clip(grads1+grads2), which makes more sense for me.
I haven't run the…
0HaNC updated
4 years ago
-
Click to expand!
### Issue Type
Bug
### Have you reproduced the bug with TF nightly?
Yes
### Source
source
### Tensorflow Version
v2.12.0-rc1-12-g0db597d0d75 2.12.0
### C…
-
I try to full finetune the deepseek-vl on multi-nodes with deepspeed-zero2. The loss and acc work fine in the beginning: nomral steps vary from the number of used nodes. However, the training loss and…
-
To make sure the grads are calculated correctly in different layers.
I'm just start reading the code and I think if we have a "Gradient Check", is a good for ensuring the grads are calculated corre…
-
### 🐛 Describe the bug
There is an issue with tracing hooks when the hook contains list append operations (it also depends on which element is added; for example, adding an integer causes no issue). …
-
Hi,
Here is a simple example in which check_grads produces an error :
```
from autograd.extend import primitive, check_grads, defvjp, defjvp
@primitive
def f(x):
return x**2
def f_jvp_…