If you see this, you were blocked because you misused memory and the rest of this was poor quality tokens
@swyx
-
Timeline sentiment shift on GPT5 mirrors o1 pattern
By
–
watching the timeline flip on GPT5 sentiment from negative to positive is pretty funny for those too cooked by the algorithm to remember, the same thing happened to o1
-
Effectiveness of System Prompting for Performance Optimization
By
–
does system prompting for juice really.. work? lol?
-
GPT-5 Visual Reasoning Capabilities Show Weakness in Testing
By
–
gpt5 visual reasoning is the weakest so far in my testing
-
Overwhelmed by Important Papers: The Peter Principle of Research
By
–
i know its ironic that its me saying this but i am feeling very overwhelmed by the number of impt papers i havent yet made the time to read haha. like theres a Peter Principle of papers where once youve read enough to do something interesting you are then given so much to do you
-
Stable Diffusion Changed My Life and Career Direction
By
–
stable diffusion changed my life. majorly updated me in a way nobody else did, and enabled me to drop everything i was doing. will always be grateful.
-
Character-Level Tokenization Success in DNA Language Models
By
–
watch me realize in realtime that @gdb is deadly serious about character-level tokenization and it's already working well for @pdhsu, @SKonermann, and @patrickc's @ArcInstitute 's 40B model that is the "GPT-1 of DNA"
— swyx 🐣 (@swyx) 15 août 2025
with biological language models (instead of natural language),… https://t.co/SMA81JYX0b pic.twitter.com/XXsMmDEZWewatch me realize in realtime that @gdb is deadly serious about character-level tokenization and it's already working well for @pdhsu
, @SKonermann
, and @patrickc
's @ArcInstitute 's 40B model that is the "GPT-1 of DNA" with biological language models (instead of natural language), -
ChatGPT Memory Falls Short on Format Preferences
By
–
my criticism of chatgpt memory is it doesnt bother to remember the structure and format i associate with a corporate timeline and instead just wants to store that i like pineapples or whatever come on guys step it up
-
Uncertainty About GPT-5-Thinking Development Work and Improvements
By
–
we still dont know if they did extra work (in the last 4 months since o3/o4mini release) on the reasoner they call gpt-5-thinking or if they just packed up the existing o4 they had lying around internally and shipped it the very slight improvements in benchmarks suggests the
-
Latent Space Podcast: GPT-5, StarGate, AI Engineers & More
By
–
gdb ep @latentspacepod we asked all your burning questions on gpt5, gptoss, stargate, RL, $300m ai engineers and more! thank you @lindsmccallum for helping us capstone the biggest launch week of the year*! (*until devday )