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
Is your feature request related to a problem? Please describe.
I'd like to get quick access to use Phoenix's awesome UMAP embedding visualization for embedding vectors from a specific span name. For example, I have a RAG application that is misbehaving. I peek at embedding vectors with span name "user-input" for a quick spot-check for outliers or mode collapse. If that passes the smell test, I could do the same with "retrieved-documents". etc. etc.
Describe the solution you'd like
Not sure the best UI/UX implementation here, but it would be nice to say, filter by a span name, and then one-click visualize all embeddings in that name space.
Describe alternatives you've considered
The only way to achieve this access and visualization currently is to query the spans to a DataFrame and then import the DataFrame as a dataset with a schema.
Additional context
This is a nice-to-have. but I think it would round out Phoenix's power in terms of complete agent observability.
The text was updated successfully, but these errors were encountered:
Is your feature request related to a problem? Please describe.
I'd like to get quick access to use Phoenix's awesome UMAP embedding visualization for embedding vectors from a specific span name. For example, I have a RAG application that is misbehaving. I peek at embedding vectors with span name "user-input" for a quick spot-check for outliers or mode collapse. If that passes the smell test, I could do the same with "retrieved-documents". etc. etc.
Describe the solution you'd like
Not sure the best UI/UX implementation here, but it would be nice to say, filter by a span name, and then one-click visualize all embeddings in that name space.
Describe alternatives you've considered
The only way to achieve this access and visualization currently is to query the spans to a DataFrame and then import the DataFrame as a dataset with a schema.
Additional context
This is a nice-to-have. but I think it would round out Phoenix's power in terms of complete agent observability.
The text was updated successfully, but these errors were encountered: