Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

fix(ai): improve AI selection algorithm #3030

Merged
merged 2 commits into from May 2, 2024
Merged

fix(ai): improve AI selection algorithm #3030

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
5a08c6d
fix(ai): improve selection algorithm
rickstaa Apr 23, 2024
a3007cf
refactor(ai): enhance selection algorithm retry logic
rickstaa Apr 24, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
12 changes: 10 additions & 2 deletions server/ai_process.go
View file
Open in desktop
Expand Up @@ -12,6 +12,7 @@ import (
"net/http"
"path/filepath"
"strings"
"time"

"github.com/livepeer/ai-worker/worker"
"github.com/livepeer/go-livepeer/clog"
Expand All @@ -21,7 +22,7 @@ import (
"github.com/livepeer/lpms/stream"
)

const maxProcessingRetries = 4
const processingRetryTimeout = 2 * time.Second
const defaultTextToImageModelID = "stabilityai/sdxl-turbo"
const defaultImageToImageModelID = "stabilityai/sdxl-turbo"
const defaultImageToVideoModelID = "stabilityai/stable-video-diffusion-img2vid-xt"
Expand Down Expand Up @@ -311,8 +312,11 @@ func processAIRequest(ctx context.Context, params aiRequestParams, req interface

var resp *worker.ImageResponse

cctx, cancel := context.WithTimeout(context.Background(), processingRetryTimeout)
rickstaa marked this conversation as resolved.
Show resolved Hide resolved
defer cancel()

tries := 0
for tries < maxProcessingRetries {
for {
tries++

sess, err := params.sessManager.Select(ctx, cap, modelID)
Expand All @@ -334,6 +338,10 @@ func processAIRequest(ctx context.Context, params aiRequestParams, req interface
clog.Infof(ctx, "Error submitting request cap=%v modelID=%v try=%v orch=%v err=%v", cap, modelID, tries, sess.Transcoder(), err)

params.sessManager.Remove(ctx, sess)

if cctx.Err() == context.DeadlineExceeded {
rickstaa marked this conversation as resolved.
Show resolved Hide resolved
break
}
}

if resp == nil {
Expand Down