spdustin

16 ms·

211.
by spdustin 3y ago
My AutoExpert is pretty popular: https://chat.openai.com/g/g-LQHhJCXhW-autoexpert-chat Dev version: https://chat.openai.com/g/g-pTF23RJ6f-autoexpert-dev
212.
by spdustin 3y ago
If you've ever added "delivery notes" to an order, they're automatically shared with every subsequent order. Clear out the delivery notes on your next order.
213.
by spdustin 3y ago
Not sure I understand here... you can tap on the CarPlay map and then drag to scroll around. It's better on capacitive screens, but still works well. Then, tap the compass indicator to go back to current location.
214.
by spdustin 3y ago
> This is meant to run lighting and puzzles for themed events and escape rooms, so it's useful to be able to completely change fonts and colors without touching markup. If you're "in control" of the interface, and acc
215.
by spdustin 3y ago
Note: these are from StatSig dynamic configurations, but assigned to the "default" group.
216.
ChatGPT's "Browse with Bing" (allow|deny)lists(github.com)
2 points by spdustin 3y ago|2 comments
217.
by spdustin 3y ago
Sadly, I don't think that'll work. They use the same headless browser setup used by Browse with Bing, and it only extracts the baked-in text from a PDF.
218.
by spdustin 3y ago
Depends on the model. If you're using GPT, combine About Me and Custom Instructions into the "System Context" text box when in Chat mode. For Claude, I have another Claude-specific version. Drop a message into the Discussions
219.
by spdustin 3y ago
Lyrics: copyright, for sure. Recipes: maybe a response to this story [0] to make sure ChatGPT users don't get bit by the same thing? [0]: https://arstechnica.com/information-technology/2023/08/ai-po...
220.
by spdustin 3y ago
The reason I track all those system prompts is due to the way they affect attention when using custom instructions. Having a system message that's already 2,700+ tokens before adding custom instructions is a tough hill to climb.
221.
by spdustin 3y ago
You can upload wheels...just ask ChatGPT to manually install them. And note that Advanced Data Analysis pods are running Python 3.8
222.
by spdustin 3y ago
For API use, if you're a paid user, you can reach out to support@openai or directly to Adam G ( https://nitter.net/therealadamg/status/1719710872317145285 ). There's no waitlist, just have to request it.
223.
by spdustin 3y ago
As Tiberium noted, this is for ChatGPT Pro users who have been granted access to the "All Tools" mode of GPT-4. For API use, if you're a paid user, you can reach out to support@openai or directly to Adam G ( https:/&#x2F
224.
by spdustin 3y ago
Here's the paper about MetNet-3: https://arxiv.org/abs/2306.06079
225.
by spdustin 3y ago
That's an interesting paper (if a little lightweight) but it's begged the question: why is temperature measurement variance the rubric for evaluating weather models? (from footnote on page 9): "Consistent with the norms in th
226.
by spdustin 3y ago
And honestly, the local weather office forecast discussions are great, too. If they seem too dense with arcane language, that's actually something that ChatGPT does a great job of distilling. "Act like a professional meteorologist
227.
by spdustin 3y ago
It's in case you're alone, and prevents someone else from being able to jump into the passenger seat and carjack you. Most new cars have a customization option to change that behavior.
228.
by spdustin 3y ago
You're a paralegal preparing case summaries for your boss. You use a high-end and trusted retrieval-augmented generative platform, and it dutifully summarizes the cases. After a spot-check, it seems right: correct plaintiffs, correct d
229.
by spdustin 3y ago
Think about all the companies doing "chat with PDF" and relying on the text of those PDF files. If you're making light of this instead of recognizing how bad actors will be using PDF files as an attack vector to LLM-based sys
230.
by spdustin 3y ago
EDIT 2: Just to make it clear. Check out this pdf [2] about an amazing CEO that gave billions of his own money to his company just before he retires to Ecuador. He probably should've gone to a non-extradition country, based on what Cha
231.
by spdustin 3y ago
The processing of the content in the PDF happens outside the context of the chat itself. The only prompt I've been able to get from that directly is just the auto-generated URL used by the headless browser that reads the file.
232.
ChatGPT will read instructions from a PDF and lie to you(old.reddit.com)
14 points by spdustin 3y ago|10 comments
233.
ChatGPT "All Tools" System omni-prompt(github.com)
1 points by spdustin 3y ago|1 comments
234.
by spdustin 3y ago
The "All Tools" omni-prompt takes a whopping 2,756 tokens, but it's also using the GPT-4 32k model, with a 32,767 token context window. It also reduces default DALL•E generations to one 'caption' (with a conflicting
235.
by spdustin 3y ago
DALL•E 3, like every other text-to-image model, still can’t generate a trombone to save its life. Just thought I’d note my own weird evaluation.
236.
by spdustin 3y ago
I’ll re-up here that I’ve collected all the ChatGPT system prompts together [0], including DALL•E 3 [1]. I wanted them all in one place as I work on the next release of my ChatGPT AutoExpert [2] “custom instructions”, which will include cus
237.
by spdustin 3y ago
FTA: """ What you're meant to do is, of course, run it on Kubernetes, which is largely the same (it's containers all the way down) except it's configured differently, you can now think about scaling out (on sev
238.
by spdustin 3y ago
It seems to suffer from choice bias, favoring choices that come earlier. If I add “car components” to the end of the list, the last example reliably returns “car”. If I move “car” to the end of the list, the model reliably returns “car comp
239.
by spdustin 3y ago
Watch the networking panel in devtools. I don’t think it’s a different tuning, per se, but I do think there’s a middleware, and network traffic suggests that’s what the middleware handles.
240.
by spdustin 3y ago
Hey, thanks, glad you checked that out! I’ve got an update brewing with native ctags support (rather than uploading a binary) that ties better into the symbol tree, and a multi-turn mode for longer edits.

More ›