Francisco Herrera

Results 11 comments of Francisco Herrera

sounds like EXO is trying to only use VRAM and do inference on the GPU instead of hybrid inference or resorting to CPU-only inference, i ran into the same issue...

> I think I have same issue. I am trying to run a `Llama 3.2 1B` on Arch Linux 16 GB of ram Intel CPU. I got same error `Out...

I thought about implementing a docker workaround too it's reliable and effective but I believe it is not ideal since it implies that a network is also emulated in the...

I too am interested in the powerinfer 2 source code

I found this: https://x.com/hodlenx/status/1800788808272937297 they said they were working on open sourcing it edit after reading the powerinfer paper i believe the fastest way to open source it would be...

> > I met a problem, which seems to be related to sockets. How to solve it? > > (exo) root@localhost:~/exo/exo# CLANG=1 python main.py Traceback (most recent call last): File...

if i open termux and then start proot with " PROOT_NO_SECCOMP=1 pd login exo ", then try to run exo by running " python3 main.py " in the folder where...

Hello according to this https://www.reddit.com/r/termux/comments/1dyvfgq/cannot_bind_netlink_socket_permission_denied/ if you see permission denied you will need to root your device otherwise it won't work. A workaround could be to have the device communicate...

Not right now as this project currently is optimized for a few transformer ai model architectures and deepseek R1 does not use one of them. I believe it would be...

maybe the frequencies are somewhat off or the bandwidth needs to be set manually?