Charles Loder
@charles_loder
Making the internet a more Hebrew friendly place
Vous pourriez aimer
Looks like Gemini 3 isn't available via API key! At least for the free tier. A shame cause a quick test in their studio showed it doing what I needed perfectly
I know I'm gonna get hit up with a bunch of bot responses, but… What some are good #ai blogs I should be following as a developer?
This idea has been rattling in my head. Instead of focusing on how "intelligent" a model is, the real power of LLMs shines when their intelligence is constrained by limiting what data the model can draw on. Not sure if I'm missing anything in this graph
Had to make an account just to try out a sandboxed version of an API, and on top of that you HAVE to subscribe to the newsletter
One of the best parts of being unemployed is that I now get how people develop anxiety from constantly checking their email
You can now run OCR over Yiddish archival pages with Qwen 3 VL! **Take the CER and WER with a grain of salt. The test data mimics the training data very closely. This should work very well for clean docs, but the big benefit of a VLM, that knowledge will transfer to messy…
In true dev fashion, I may be over engineering the job search
The job hunt continues… If anyone knows of any open developer roles, preferably in ed tech or ed tech adjacent, I'd appreciate the lead! If you don't, please repost!
Checkout Built-in AI from @hoeg_jakob for running models on device and directly in the browser using the Vercel AI SDK 🎉 github.com/jakobhoeg/buil…
> Online trends tend to start from chaos and gain momentum without explanation: For example, “404 Not Found” went from a nonsense phrase found on a dead-end page of the web Ummm…404 is an HTTP status code, not a nonsense phrase. Even AI knows this hollywoodreporter.com/tv/tv-features…
Small models work great for GLAM but there aren't enough examples. With @wjb_mattingly I'm launching small-models-for-glam on @huggingface to create and curate models that run on modest hardware and address GLAM use cases. More coming soon. Follow the org to keep up to date!
Multimodal chat with visions models in the browser is surprisingly difficult. I've tried Transformers.js, web-llm, and MediaPipe, and they all seem to have some critical shortcoming. Am I doing this wrong? Or anyone have recommendations?
United States Tendances
- 1. Cheney 42.2K posts
- 2. Nano Banana Pro 14.9K posts
- 3. First Take 43.5K posts
- 4. #AcousticPianoSnowGlobe 1,957 posts
- 5. Cam Newton 2,220 posts
- 6. Stephen A 36.8K posts
- 7. SEDITIOUS BEHAVIOR 12.2K posts
- 8. FINAL DRAFT FINAL LOVE 362K posts
- 9. Trump and Vance 29.4K posts
- 10. #LoveDesignFinalEP 328K posts
- 11. #TSTheErasTour 1,606 posts
- 12. #XboxPartnerPreview 1,407 posts
- 13. Bush 51.9K posts
- 14. Sedition 86.3K posts
- 15. Husqvarna 1,077 posts
- 16. Treason 56.1K posts
- 17. Godzilla 21.9K posts
- 18. Stuart Scott 1,942 posts
- 19. Nnamdi Kanu 176K posts
- 20. #WeekndTourLeaks 1,097 posts
Vous pourriez aimer
-
Jill Firth
@jilfirth -
Tyndale Bulletin
@TyndaleBulletin -
Peter Gurry
@pjgurry -
Michelle Knight
@mlizknight -
Fresch
@cjfresch -
John Meade
@drjohnmeade -
Dirk Jongkind
@DJngKnd -
Chip Hardy
@drchiphardy -
Charlie Trimm
@charlie_trimm -
Timothy (Tim) Lee
@Timothy_A_Lee -
HebrewDaybyDay hosted by Gary Edward Schnittjer
@HebrewDaybyDay -
Seth M. Ehorn
@smehorn -
Roberto Carrera
@R28Carrera -
Brent Niedergall
@BrentNiedergall -
Jacob Cerone
@ceronej
Something went wrong.
Something went wrong.