hellovai

6 ms·

31.
by hellovai 2y ago
Hey everyone! One of the creators of BAML here! Appreciate sharing this post. For anyone interested in playing around with an interactive version of BAML online, check it out here: https://www.promptfiddle.com
32.
Prompts are using 4x more tokens than you need(boundaryml.com)
7 points by hellovai 3y ago|2 comments
33.
by hellovai 3y ago
All we wanted was our prompts in our codebase, not on some website. Then just to see the prompts before we ran the code. Then comments inside of prompts. Then just less strings everywhere, and more type-safety... https://imgs.xkc
34.
Show HN: BAML - A new programming language to get structured data from LLMs(github.com)
18 points by hellovai 3y ago|2 comments
35.
by hellovai 3y ago
Have you seen arcs search? Its one of my favorite feature where cmd + f seamlessly transitions to AI chat when it doesn't find the text directly. seems like a +1 if cmd + f seamlessly transitions from direct text -> ai chat -> se
36.
by hellovai 3y ago
oh interesting, this is gonna be pretty helpful if it could actually recognize some pattern of errors even if the stack trace is different. (i.e. some shared function call fails, but that function is used within a bunch of different functio
37.
by hellovai 3y ago
I appreciate the feedback, and also agree that chatgpt is a no-go for many use cases. We're working on putting together a better comparison specifically along the lines of accuracy between the LLMs (chatgpt, bard, falcon) and also trad
38.
by hellovai 3y ago
That's a great summary and insight. We should likely use that verbiage to help make it more crystal clear :)
39.
by hellovai 3y ago
That's very interesting! What sort of direction did you head in with prompt optimization? Was it mostly in shrinking it and then using multi-shot examples? We found that shorter prompts (empirically) perform better than longer prompts.
40.
by hellovai 3y ago
SetFit is fairly good, and we do help train SetFit like models for the results you get, however, the issue with SetFit is that its latency and cost benefits come at the price of flexibility. If you want to add a new class, to update an exis
41.
by hellovai 3y ago
Thanks Alex, in this article we focused more on deployment comparisons, for example the cost and latency of what it would take to deploy a BERT based model vs LLMs. In a future article, we're planning on posting accuracy comparisons as
42.
by hellovai 3y ago
oh! Good catch! Fixed this, and will update in the release.
43.
by hellovai 3y ago
That's a good point! We're actually working on integrating this as well, but in practice, what we've found is that LLM's in general don't like to respond with empty strings for example. My hypothesis here is that du
44.
Classifying customer messages with LLMs vs traditional ML(trygloo.com)
261 points by hellovai 3y ago|123 comments