-
I noticed that the training code did not update the ema parameters( ema_model.step(model.parameters()) ). Then I have checked the checkpoint, I found ema parameters have updated. I'm not sure, where i…
-
### Describe the Bug with repro steps
1. Create new Logic Apps in West US 2.
2. Switch new UI Logic Apps designer.
3. Choose a action such as Update Device Properties of Azure IoT Centaral V3 conn…
-
hello
@LTH14
I am re-training MAR on Imagenet Dataset, and evaluate checkpoint on the epoch 1. However, the image sampled from epoch-1 is black. I want to know why?Does it means the epoch I used i…
-
I am using my self-trained autoencoder as the encoder to train on the CIFAR-10 dataset. After 500 epochs, the loss dropped to around 0.1, but the reconstructed images are almost all white, with pixel …
-
Thank you for your open-source work. I would like to ask you some questions. I tried to use diffloss for a speech generation task, adopting the next token prediction approach. This corresponds to the …
-
想问一下这里是如何实现ema模型的保存呢:
![image](https://github.com/user-attachments/assets/654c4a13-d10a-4fb1-bc0c-0570d5a2eed8)
从这里没有看到model被显示的修改,是deepspeed内完成的么?
我有check保存下来的model和ema_model,发现参数是完全一样的,所以不确定是不是…
-
## 🚀 Feature
How about add EMA as callback?
### Motivation
I have had difficulty in applying ema. I think it would be nice if there are EMA as callback.
### Pitch
If user add ema …
-
# Modifying parameters of FSDP-wrapped module by hand without summon_full_params context
## Issue description
I am training a large language module using FSDP.
I want to store EMA weights wh…
-
From this paper: https://proceedings.systemdynamics.org/2016/proceed/papers/P1311.pdf
> From this large toolbox of Python modules mesa was used to implement the ABM (Masad and Kazil, 2015) and PyS…
-
Hi,
Shouldn't this line
https://github.com/joeldg/bowhead/blob/ee67a6ef0a6f82100e1ac7940a4dad172a9a8bd1/app/Traits/CandleStrategies.php#L24
be changed to
requires $ema = trader_ema($dat…