I guess CLIP is another example of this, where the two modes are text and binary images
@simonw
-
Voyage AI Input Types for Embeddings Documentation
By
–
Thanks, yeah it looks like they call these "input types" https://
docs.voyageai.com/docs/embeddings -
RAG Implementation: Embedding Passages and Queries Separately
By
–
Being able to embed your content as "passage" but questions people ask about it as "query" is useful for implementing RAG – a user's question might not naturally embed to a similar location as content that answers that question, this trick helps fix that
-
Embedding Models with Content Mode Separation
By
–
What embedding models are there with separation between different modes of content? I know E5-Large-V2 has that ("passage" vs "query"), and @nomic_ai have "search_query", "search_document", "clustering", "classification" https://
docs.nomic.ai/reference/endp
oints/nomic-embed-text
… Any other good examples? -
Challenges extracting table data from screenshot images
By
–
Definitely founded – I've had particular trouble getting table data out of screenshots of tables, but I don't trust it very much at all yet
-
Where to Find AI Model Output Token Limits Documentation
By
–
Do you know where they document their output token limit? I can't find that for any of their models
-
Pushing the limits of AI system capabilities and risks
By
–
Yes, that's a very real risk. I'm starting to try and push the limits of what makes sense to pipe through these things
-
Data Extraction Usability: Addressing Truncation Problems for Users
By
–
The bigger problem here is a usability one: explaining to end users why their extracted data was randomly cut off half way through (or risking them not noticing) isn't great
-
Documentation Token Limit Testing and Performance Analysis
By
–
The documentation says it should cut off at 4096 tokens of output, but I haven't stress tested it myself yet
-
Text extraction challenges with token output limitations
By
–
The HTML thing was really just an illustrative example – the general challenge is that there are plenty of text extraction tasks where the output is > 8196 tokens so the more output tokens we can have the easier these things are to put into practice