I am trying to get accelerate working on a video task and I am running into problems with processes getting stuck. Here’s a brief summary of my problem: I have multiple directories containing multiple (up to a thousand) image frames. Because loading all images for a batch of videos at once is not p…

Problems with hanging process at the end when using dataloaders on each process

varox34 January 1, 2025, 10:34am 6

I am facing same issue but with LLMs I am inferencing Llama2 model using dataloaders and process gets stuck in the 2-gpu setup

1 Like

Topic		Replies	Views
Accelerate stalls when using Tensor Dataset 🤗Accelerate	0	317	December 31, 2023
Accelerate Distributed Randomly Hangs 🤗Accelerate	0	99	September 11, 2024
HF Accelerate uses multiple GPUs even when setting `num_processes` to 1 🤗Accelerate	0	91	August 2, 2024
Error occurs when loading additional parameters in multi-gpu training Beginners	1	349	December 14, 2021
Accelerate test stuck on training 🤗Accelerate	2	2377	January 24, 2024