Twake Drive – An open-source alternative to Google Drive
github.com
Mesh2Motion – Open-source web application to animate 3D models
mesh2motion.org
Why formalize mathematics – more than catching errors
rkirov.github.io
A sharded DuckDB on 63 nodes runs 1T row aggregation challenge in 5 sec
gizmodata.com
The key to universe exists may lie in an 1800s knot idea science once dismissed
phys.org
Where's the AI Design Renaissance?
learnui.design
ChunkLLM: A Lightweight Pluggable Framework for Accelerating LLMs Inference
arxiv.org
'Attention is all you need' coauthor says he's 'absolutely sick' of transformers
venturebeat.com
Linux disk I/O diagram (2024)
zenodo.org
Ivy League psychologist: 'Bring your whole self to work' is bad advice
cnbc.com
Asahi Linux Still Working on Apple M3 Support, M1n1 Bootloader Going Rust
phoronix.com
Padlet (YC W13) Is Hiring in San Francisco and Singapore
padlet.jobs
Iceland just found its first mosquitoes
cnn.com
Counter-Strike's player economy is in a multi-billion dollar freefall
polygon.com
Why can't transformers learn multiplication?
arxiv.org
LightlyStudio – an open-source multimodal data curation and labeling tool
github.com
Public Montessori programs strengthen learning outcomes at lower costs: study
phys.org
Poker fraud used X-ray tables, high-tech glasses and NBA players
bbc.com
/dev/null is an ACID compliant database
jyu.dev
This is awesome! I wonder what the benefits and challenges of doing this at the AST level vs in LLVM IR are.
How much of the language needs to be supported by the AST bytecode interpreter to support precomputing constexprs?
If information about constexprs could be propagated to LLVM IR, could and LLVM IR interpreter be used to precompute constexprs?
Also, would a complete AST bytecode interpreter be useful for checks in clang static analyzer?