-
I merged a mistal 8x7b model with the lora adapter, and I save the .pt with torch.save(model.state_dict(), 'path_to_model.pt')
However, when I use vllm to inference on the new merged model, I fai…
-
### What happened + What you expected to happen
When running the script below, the memory utilization of the `Supervisor` actor increases slowly over time. This usually leads to OOM for more complex …
-
### Search before asking
- [X] I have searched the Ultralytics YOLO [issues](https://github.com/ultralytics/ultralytics/issues) and [discussions](https://github.com/ultralytics/ultralytics/discussion…
-
The feature of max download workers with parsing account SN seems to be working fine.
I would not download any song if one of the parameter changes.
-
Hi there,
I implemented edgetunnel using rust, welcome to try to develop and use.
-
As of today, workers can select which scrapers they want to run.
But at a central level, we have no control over which scrapers are allowed to run which scrapers, and this could be an issue for var…
-
Currently, disk size assigned to each workers is not much used AFAIK.
However, the disk size configured in zimfarm.config is displayed in the UI.
Would be great to at least check if the disk siz…
-
-
Hello,
I started bjoern server with 2 workers and sent 6 requests in parallel - scripts below
the server just sleeps for 1 second and then responds
What I would expect to happen is that the r…
Pebin updated
3 months ago
-
When using CSP (Content Security Policy), inline workers require the use of the script-src blob: policy which is a big no no and opens up the whole website for js injection.
Would/is it be possible t…