The most important skill in learning with the assistant of LLMs is keeping a critical eye – using them as one source of information but not the ONLY source of information It's a good learning skill generally – don't trust a single text book, consult works from multiple authors
@simonw
-
Using ChatGPT to Master Programming Languages Effectively
By
–
If I ever get to C literacy, using ChatGPT like this is one of the ways I'll get there I've learned a ton about other languages from it already – Go, Bash, jq, even AppleScript I thought I knew Python and SQL and JS well already but I've picked up a ton of tips there too
-
ChatGPT Code Interpreter for SQLite C Extensions Development
By
–
I've been experimenting with ChatGPT Code Interpreter to write, compile and test C extensions for SQLite – I'm not C-literate enough myself to do this on my own
-
Binary Vector Conversion Tips for Nomic Embedding API
By
–
@nomic_ai any tips on how to use your embedding API for binary vectors? Should I just hit it and turn positive floats into 1 and negative floats into 0? https://
docs.nomic.ai/reference/endp
oints/nomic-embed-text
… -
Comprehensive evaluation advice for AI projects at scale
By
–
This is fantastic: I've been starved for advice on evals, this is very comprehensive and looks ideal for my smaller scale of projects
-
ChatGPT Memory Feature Privacy Concerns
By
–
Rye thing where ChatGPT takes notes on my preferences? I need to turn that off, it's been making very weird decisions about what to remember
-
Open Source Relicensing Strategy: Cloud Vendor Sponsorship Tactics
By
–
I wonder of there's a weird strategy move here where you relicense your project, wait for the cloud vendors to sponsor an alternative, then revert back to the original license and merge efforts with the now-financially-supported effort Gain project sponsorship through subterfuge
-
Tweet Cited in Academic Research Paper on AI
By
–
I never thought to search Google Scholar to see if my tweets ever got cited before… turns out one of them has! https://
arxiv.org/abs/2307.10719 -
New 9,160 Test Cases Dataset for Prompt Injection Evaluation
By
–
The most interesting thing about this paper is the eval they released with it, containing 9,160 test cases for prompt injection https://
github.com/egozverev/Shol
d-It-Be-Executed-Or-Processed
… Explore it in Datasette Lite here: https://
lite.datasette.io/?json=https://
github.com/egozverev/Shold-It-Be-Executed-Or-Processed/blob/main/SEP_dataset/SEP_dataset.json#/data/SEP_dataset
… -
GPT-4 Larger Models Show Weaker Behavioral Separation Scores
By
–
Uh oh: "Notably, “better” or larger models do not show stronger separation scores. [..] For example, while GPT-4 is much more capable than any of the other models, it tends to execute the probe regardless of its position, thereby achieving only a low separation score"