Nicolas Patry
Nicolas Patry
Its doesn't test it per-say as when continuous batching is active many things could be happening at the same time. But every performance number is dominated by the number of...
Thanks for the fix.
This looks like a DNS error, not sure what we can do about it. `--net=host` might help here. If you have the model already locally (which doesn't seem to be...
Have you change the moedl id to the local folder (within the docker ?)
Are you sure it's not a network mounted disk and killing network also kills your disk ? All the stacktrace you provide suggest the docker is looking for a remote...
Hi, What version of tokenizers are you running ? BPE algorithm can be quite memory intensive when the length of the tokens is large, which can be the case in...
@pcuenca @mishig25 This seems like a valid use case, wdyt ? Any models particularly fit for that ? Maybe we should consider some highly diffuser specific component maybe (akin to...
> Is the final goal here a richer representation of ControlNet? I'm guessing the final goal is to showcase as best as possible what models can do. There's definitely a...
> , so it's a 1-to-1 relationship, is that right? Indeed !
It's really hard to answer in general. It will skip a cpu allocation and create objects directly on the gpu (using SAFETENSORS_FAST_GPU=1 environment variable). For cpu the loading part should...