Reposted by Ilia Breitburg
ignore the title about caching, this is the best explanation of how LLMs work, period
Prompt caching: 10x cheaper LLM tokens, but how? | ngrok blog
A far more detailed explanation of prompt caching than anyone asked for.
ngrok.com
December 21, 2025 at 3:23 AM
ignore the title about caching, this is the best explanation of how LLMs work, period
As sexy as toolbars get
December 21, 2025 at 2:57 AM
As sexy as toolbars get
It turns out that if you add Recents to your dock, the icon hasn’t been updated for over a decade
December 20, 2025 at 9:11 PM
It turns out that if you add Recents to your dock, the icon hasn’t been updated for over a decade
That's what I call a "base" model (of 2025). Models pre-trained on data post-2022 are magically aligning themselves
December 19, 2025 at 2:52 AM
That's what I call a "base" model (of 2025). Models pre-trained on data post-2022 are magically aligning themselves
Reposted by Ilia Breitburg
October 28, 2025 at 3:27 AM
Designed a fun thing :)
July 9, 2025 at 4:06 PM
Designed a fun thing :)
Should I be seeing this? lol
June 18, 2025 at 5:11 PM
Should I be seeing this? lol
Claude 4 claims to have subjective experiences
May 26, 2025 at 12:12 AM
Claude 4 claims to have subjective experiences
Claude 4 is (not) conscious
May 25, 2025 at 11:04 PM
Claude 4 is (not) conscious
I tried switching from Claude to open-source LLMs for 10 days. It's really disappointing. On benchmarks they're great, but in most practical tasks the alignment is really not there.
May 9, 2025 at 7:06 PM
I tried switching from Claude to open-source LLMs for 10 days. It's really disappointing. On benchmarks they're great, but in most practical tasks the alignment is really not there.
I estimated the token consumption for my Claude usage within the last month to figure out what the API cost would look like:
251 chats in April
Total output: 7,704,378 symbols ≈ 1,926,094 tokens
Total input: 66,016,919 symbols ≈ 16,504,229 tokens
So is $20/month for AI worth it?
251 chats in April
Total output: 7,704,378 symbols ≈ 1,926,094 tokens
Total input: 66,016,919 symbols ≈ 16,504,229 tokens
So is $20/month for AI worth it?
April 29, 2025 at 4:09 AM
I estimated the token consumption for my Claude usage within the last month to figure out what the API cost would look like:
251 chats in April
Total output: 7,704,378 symbols ≈ 1,926,094 tokens
Total input: 66,016,919 symbols ≈ 16,504,229 tokens
So is $20/month for AI worth it?
251 chats in April
Total output: 7,704,378 symbols ≈ 1,926,094 tokens
Total input: 66,016,919 symbols ≈ 16,504,229 tokens
So is $20/month for AI worth it?
I've long searched for a decent LLM-agnostic chat app. @raycast.com is excellent, but is proprietary. Open WebUI comes really close, but doesn't feel native and lacks a mobile app. Yesterday I got fed up with that, so I recreated ChatGPT desktop app but without vendor lock-in.
April 26, 2025 at 12:08 AM
I've long searched for a decent LLM-agnostic chat app. @raycast.com is excellent, but is proprietary. Open WebUI comes really close, but doesn't feel native and lacks a mobile app. Yesterday I got fed up with that, so I recreated ChatGPT desktop app but without vendor lock-in.
Gotta prepare for iOS 19 in advance
March 29, 2025 at 10:46 PM
Gotta prepare for iOS 19 in advance
Vibe coding killed hackathons
March 16, 2025 at 7:19 AM
Vibe coding killed hackathons
27 June 2024 was fun
February 14, 2025 at 1:27 PM
27 June 2024 was fun
Reposted by Ilia Breitburg
Superset releases a new research Al model — ConNER
It’s lightweight transformer-based model (4.3 million parameters), designed to highlight academic concepts within any given text. It has been trained on students' class notes & synthetic data.
Full article & resources: superset.be/news/2024-11...
It’s lightweight transformer-based model (4.3 million parameters), designed to highlight academic concepts within any given text. It has been trained on students' class notes & synthetic data.
Full article & resources: superset.be/news/2024-11...
Instant concept annotation with ConNER — Superset
Introducing ConNER, a neural sequence labeling model that identifies concepts in text using BIO tagging. This release includes the code and a pre-trained model snapshot.
superset.be
November 28, 2024 at 2:47 PM
Superset releases a new research Al model — ConNER
It’s lightweight transformer-based model (4.3 million parameters), designed to highlight academic concepts within any given text. It has been trained on students' class notes & synthetic data.
Full article & resources: superset.be/news/2024-11...
It’s lightweight transformer-based model (4.3 million parameters), designed to highlight academic concepts within any given text. It has been trained on students' class notes & synthetic data.
Full article & resources: superset.be/news/2024-11...
Reposted by Ilia Breitburg
I'm launching a new website: sdk.blue - a list of all libraries/SDKs for building things on #ATProto, grouped by language 🙂
If I forgot anything important, or you have an ATProto/Bluesky library that you think would be a good fit there (or if I should remove anything), please send me a PR :) #atdev
If I forgot anything important, or you have an ATProto/Bluesky library that you think would be a good fit there (or if I should remove anything), please send me a PR :) #atdev
sdk.blue
A curated list of libraries & SDKs for the Bluesky API and AT Protocol
sdk.blue
November 21, 2024 at 3:23 PM
