DOCS-1186 AI Accelerator updates for 2.1.1 #6445
Conversation
Signed-off-by: Dj Walker-Morgan <dj.walker-morgan@enterprisedb.com>
djw-m
added
the
deploy
djw-m
changed the title
Signed-off-by: Dj Walker-Morgan <dj.walker-morgan@enterprisedb.com>
djw-m
added
deploy
|
FYI; this release includes the first pieces for an upcoming "document prep pipeline"; i.e. SQL primitives to chunk text and similar. We're dark-launching these for now so they should not be mentioned in the release notes. Interfaces are subject to change and not everything is finished yet. ... as discussed with @noahbaculi |
| @@ -21,7 +21,7 @@ The starting point for this process is creating a model. When you create a model | |||
| ```sql | |||
| select aidb.create_model( | |||
| 'my_local_ollama', | |||
| 'openai_embeddings', | |||
| 'embeddings', | |||
| '{"model":"llama3.3", "url":"http://llama.local:11434/v1/embeddings", "dimensions":8192}'::JSONB, | |||
There was a problem hiding this comment.
I believe we need to add a warning somewhere after this example about the pgvector limitation. Since it doesn't support storing over 2000 vectors. Otherwise, this example can be misleading.
There was a problem hiding this comment.
Trying to remember how I got to 8192... will bring the local box up and see about shrinking it
There was a problem hiding this comment.
The code may indicate the maximum is 8192, but there. is a non-obvious technical limitation of 2000.
|
|
||
| ## About Completions | ||
|
|
||
| Completions is a text completion model that enables use of any OpenAI API compatible text generation model. |
There was a problem hiding this comment.
Since this is an interface for a common API format I would probably drop the first half of this sentence and say something like:
Enables use of any OpenAI API compatible text generation model.
What Changed?
Updates for credentials and replacement of credentials.