You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I ran a search for anything Langfuse related before posting this. Didn't see anything.
I'm curious if anyone has messed with trying to send a session ID from Vane to Langfuse, so that it can be used in observability, specifically to group traces.
I am getting all traces from Vane, in Langfuse, because I have LiteLLM integrated with Langfuse and I use models through LiteLLM.
For some context, I use LM Studio as my model delivery layer. I generally don't access the models from LM Studio though. I use LiteLLM as a model proxy, primarily (at least initially) because of it's easy integration with Langfuse for observability. I have since learned how powerful something like LiteLLM is and I like the idea of a central solution for many things, so I am excited about many other things in LiteLLM. I am looking at LiteLLM as a sort of "Management Plane" and "Integration Layer" basically.
I have found though, that even though getting tracing to work in Langfuse was pretty easy, enabling other features like "Session" grouping, and a few other things is not so easy.
I was able to get session information going to Langfuse for Open-WebUI, but it wasn't easy. I sort of got lucky, because someone else had explored it, and created a pipeline function to send that data to a configured Langfuse instance.
I have tried a few things, to try to get a session ID from Vane to Langfuse, based on doing a bunch of web exploration. Have not had any success though.
I do have the "Optimized Search" function that someone created for Open-WebUI, which is designed to enable leveraging Vane and SearXNG for research right in Open-WebUI. I have found that it lacks several things, and doesn't seem to be as thorough and clean in terms of responses, even using the same models as Vane. When I use this though, I can get the session ID, because the flow is initiated through Open-WebUI obviously. This isn't perfect though and isn't really as good as I'd like.
So, I was hoping that by starting a discussion about it, maybe someone has already explored it, or someone smarter than me would be interested in it and might be able to establish a solution.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
I ran a search for anything Langfuse related before posting this. Didn't see anything.
I'm curious if anyone has messed with trying to send a session ID from Vane to Langfuse, so that it can be used in observability, specifically to group traces.
I am getting all traces from Vane, in Langfuse, because I have LiteLLM integrated with Langfuse and I use models through LiteLLM.
For some context, I use LM Studio as my model delivery layer. I generally don't access the models from LM Studio though. I use LiteLLM as a model proxy, primarily (at least initially) because of it's easy integration with Langfuse for observability. I have since learned how powerful something like LiteLLM is and I like the idea of a central solution for many things, so I am excited about many other things in LiteLLM. I am looking at LiteLLM as a sort of "Management Plane" and "Integration Layer" basically.
I have found though, that even though getting tracing to work in Langfuse was pretty easy, enabling other features like "Session" grouping, and a few other things is not so easy.
I was able to get session information going to Langfuse for Open-WebUI, but it wasn't easy. I sort of got lucky, because someone else had explored it, and created a pipeline function to send that data to a configured Langfuse instance.
I have tried a few things, to try to get a session ID from Vane to Langfuse, based on doing a bunch of web exploration. Have not had any success though.
I do have the "Optimized Search" function that someone created for Open-WebUI, which is designed to enable leveraging Vane and SearXNG for research right in Open-WebUI. I have found that it lacks several things, and doesn't seem to be as thorough and clean in terms of responses, even using the same models as Vane. When I use this though, I can get the session ID, because the flow is initiated through Open-WebUI obviously. This isn't perfect though and isn't really as good as I'd like.
So, I was hoping that by starting a discussion about it, maybe someone has already explored it, or someone smarter than me would be interested in it and might be able to establish a solution.
Thanks in advance!!
All reactions