Sangwoong Yoon

Fix the sampling from von-Mises-Fisher in Hypersphere

2

I found that `Hypersphere.random_von_mises_fisher` ([permlink](https://github.com/geomstats/geomstats/blob/944e1205ebbe506ce6865a56afc06d9f464bf6f1/geomstats/geometry/hypersphere.py#L454 )) method outputs an error for `dim=1`. It seems to work fine on higher dimensions though. The following example raises ValueError. ```python import numpy as...

Reconstructed images are not like the input images

2

Hi, First of all, thank you for a nice repo. I am trying to map an image `x` to a latent representation `z` and then map back to its reconstruction...

Can you explain why the energy is divided by b0?

1

In the following code for computing the (unnormalized) log probability, the network output is divided by `b0`. https://github.com/ruiqigao/recovery_likelihood/blob/c77cc0511dedcb8d9ab928438d80acb62aeca96f/model.py#L154 I wonder if there is a legitimate explanation for this division. `b0`...

What is noise_variance in SamplePoint about?

1

First of all, I am really amazed by the wonderful, neat interface of MOE. I appreciate it. I don't understand why there are noise_variance attribute in [SamplePoint](http://yelp.github.io/MOE/moe.optimal_learning.python.html?highlight=samplepoint#moe.optimal_learning.python.data_containers.SamplePoint). In other words,...

Hi, I am familiarizing myself with OpenRLHF by running some examples. I started with the SFT example provided by the doc https://openrlhf.readthedocs.io/en/latest/rl.html#supervised-fine-tuning but immediately faced the following error regarding dataset...

Gradient accumulation gives worse results when using DeepSpeed ZeRO 2

6

### System Info ```Shell - `Accelerate` version: 1.11.0 - Platform: Linux-5.4.0-42-generic-x86_64-with-glibc2.35 - `accelerate` bash location: /raid/swyoon/miniconda3/envs/tbitrl/bin/accelerate - Python version: 3.12.12 - Numpy version: 2.2.6 - PyTorch version: 2.7.1+cu126 - PyTorch...