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
Currently, Gateway tries to mimic OpenAI streaming behaviour and break the JSON into 4 characters (1 token) and return them individually to the client.
But this approach is redundant because the Gateway already has the full JSON response. So this just adds up to more chunks. Due to this, latency increases and it might also cause network bottlenecks due to unreasonable number of chunks.
The text was updated successfully, but these errors were encountered:
Currently, Gateway tries to mimic OpenAI streaming behaviour and break the JSON into 4 characters (1 token) and return them individually to the client.
But this approach is redundant because the Gateway already has the full JSON response. So this just adds up to more chunks. Due to this, latency increases and it might also cause network bottlenecks due to unreasonable number of chunks.
The text was updated successfully, but these errors were encountered: