Post AcXwFTLURe2eDCbGFc by apexnasi@hessen.social
(DIR) More posts by apexnasi@hessen.social
(DIR) Post #AcXhfj6uxH9RK4qRLU by simon@fedi.simonwillison.net
2023-12-06T19:35:33Z
0 likes, 0 repeats
Found the prompts for the excellent new @IceCubesApp image description generation feature - it's using GPT-4 vision and sending the image with the prompt:> What’s in this image? Be brief, it's for image alt description on a social network. Don't write in the first person.It sets max_tokens to 50https://github.com/Dimillian/IceCubesApp/blob/4f9e23296fa9c8abb812bc24f0f9a1ce0c86b28a/Packages/Network/Sources/Network/OpenAIClient.swift#L86-L101
(DIR) Post #AcXiKyfO1V31RRdZZY by dimillian@mastodon.social
2023-12-06T19:42:32Z
0 likes, 0 repeats
@simon @IceCubesApp indeed :)
(DIR) Post #AcXiYNnbtMOL9asNGK by julia@mathstodon.xyz
2023-12-06T19:43:03Z
0 likes, 0 repeats
@simon @IceCubesApp i love how it’s no longer necessary to write an entire essay worth of prompts to get good results
(DIR) Post #AcXingF1bfbch8zebo by emic@c.im
2023-12-06T19:48:20Z
0 likes, 0 repeats
@simon I’m not the only one who goes to peek at the IceCubes prompts 😂
(DIR) Post #AcXk5StS4YloSnwmIa by dom@chaos.social
2023-12-06T20:02:34Z
0 likes, 0 repeats
@simon Does max_token affect the answer being generated, i.e. lower max_tokens = shorter answers? Or is the response just cut off after reaching the limit?
(DIR) Post #AcXrPQeknasEzpXT28 by iwootten@mastodon.social
2023-12-06T21:24:35Z
0 likes, 0 repeats
@simon @IceCubesApp do you collate any of these found prompts anywhere on your blog?
(DIR) Post #AcXu46ZoYjWgV9kgq0 by simon@fedi.simonwillison.net
2023-12-06T21:52:22Z
0 likes, 0 repeats
@iwootten @IceCubesApp No, but I've been thinking it would be really interesting to build a collection of these somewhere - especially a collection that noticed when people changed their prompts
(DIR) Post #AcXwFTLURe2eDCbGFc by apexnasi@hessen.social
2023-12-06T19:46:37Z
0 likes, 0 repeats
@dimillian @simon @IceCubesApp will this work on a video anytime soon?
(DIR) Post #AcXwFUFr458120yH6e by simon@fedi.simonwillison.net
2023-12-06T22:18:49Z
0 likes, 0 repeats
@apexnasi @dimillian @IceCubesApp GPT-4 doesn't have the ability to analyze video yet - you can send it a bunch of static frames from a video and ask it to stick something together from those but that's both expensive, slow and unlikely to produce great results
(DIR) Post #AcXwR7osIuXwvPtvI8 by simon@fedi.simonwillison.net
2023-12-06T22:19:34Z
0 likes, 0 repeats
@dom as far as I can tell GPT-4 does somehow take the max tokens into account when deciding what to generate