- 0.9B model (GLM-OCR)
~$0.002/page
~$5 total on an L4 GPU
Before (old Tesseract ocr) → After
Read here: buff.ly/M7N4yjx
Read here: buff.ly/M7N4yjx
Read here: buff.ly/uEBxzdK
Some key findings...
Some key findings...
"For the last decade, it has been hard to stray off the beaten path of accepted wisdom that scaling training parameters drives innovation.
However, the relationship between training compute + performance is uncertain + rapidly changing."
Short thread: this is actually an engineering paper, taking as a starting points ideas already exposed in an original Hyper-Connections (HC) paper from ByteDance.
Short thread: this is actually an engineering paper, taking as a starting points ideas already exposed in an original Hyper-Connections (HC) paper from ByteDance.
tl;dr they trained a model to answer questions about another model’s thoughts
alignment.anthropic.com/2025/activat...
tl;dr they trained a model to answer questions about another model’s thoughts
alignment.anthropic.com/2025/activat...
www.theguardian.com/world/2025/d...
www.theguardian.com/world/2025/d...
It's 3,000 lines of code mostly written by coding agents over a couple of months simonwillison.net/2025/Dec/14/...
It's 3,000 lines of code mostly written by coding agents over a couple of months simonwillison.net/2025/Dec/14/...
though when compared to what's being done elsewhere, these numbers feel quite disappointing... other countries with similar solar potential have been adding capacity more than 5x faster than we have in the UK!
www.pv-magazine.com/2025/11/27/u...
though when compared to what's being done elsewhere, these numbers feel quite disappointing... other countries with similar solar potential have been adding capacity more than 5x faster than we have in the UK!
Feels like this should have been bigger news...
www.greenpeace.org.uk/news/britain...
Feels like this should have been bigger news...
www.greenpeace.org.uk/news/britain...