feature(worker): Allow configuring inference job timeout and ollama keep alive. Fixes #389 #224

2 files changed, 2 insertions, 1 deletions

diff --git a/apps/workers/inference.ts b/apps/workers/inference.ts
index fa83140f..071f4742 100644
--- a/apps/workers/inference.ts
+++ b/apps/workers/inference.ts
@@ -103,6 +103,7 @@ class OllamaInferenceClient implements InferenceClient {
       model: model,
       format: "json",
       stream: true,
+      keep_alive: serverConfig.inference.ollamaKeepAlive,
       messages: [
         { role: "user", content: prompt, images: image ? [image] : undefined },
       ],
diff --git a/apps/workers/openaiWorker.ts b/apps/workers/openaiWorker.ts
index 55695938..8bd2cf4a 100644
--- a/apps/workers/openaiWorker.ts
+++ b/apps/workers/openaiWorker.ts
@@ -81,7 +81,7 @@ export class OpenAiWorker {
       {
         concurrency: 1,
         pollIntervalMs: 1000,
-        timeoutSecs: 30,
+        timeoutSecs: serverConfig.inference.jobTimeoutSec,
       },
     );