galsapir

93 karmajoined January 17, 2026On HN

sparse thoughts on things i find online

Recent submissions

Creative Reading: Scaffolding Reading for Transformation arxiv.org
1galsapir20hdiscuss
No agents after 8 pm sparsethought.com
2galsapir12ddiscuss
AI Has Plunged the Book Publishing Industry into Utter Chaos wsj.com
24galsapir23d21 comments
Can AI agents conduct open-ended AI research?arxiv.org
4galsapir1modiscuss
Ask HN: Any Good Pi.dev Setups?
3galsapir1mo1 comment
Giving a domain a hill to climb: benchmarking as data activation sparsethought.com
12galsapir2mo7 comments
A bitter lesson for medicine, or a benchmark problem?sparsethought.com
2galsapir2modiscuss
Can LLMs Beat Classical Hyperparameter Optimization Algorithms?arxiv.org
120galsapir3mo20 comments
Gemma 4 E4B as a primary local LLM (replaced Qwen)digg.com
2galsapir3modiscuss
PEEK: Give Your Agent an Orientation Cache (MIT CSAIL, Khattab group)zhuohangu.github.io
3galsapir3modiscuss
Hyperagents (Meta Research)arxiv.org
2galsapir3modiscuss
The Unreasonable Effectiveness of HTML claude.com
3galsapir3mo2 comments
The Comparator in Clinical AI sparsethought.com
2galsapir4mo1 comment
Borges' cartographers and the tacit skill of reading LM output galsapir.github.io
40galsapir5mo10 comments
Best read of 2026 so far was written in 1880 galsapir.github.io
1galsapir6mo1 comment
Anthropic launched community ambassador program claude.com
1galsapir6modiscuss
LLMs as nudging research towards luke-warm middle nature.com
1galsapir6modiscuss
How do you evaluate a foundation model before you know what it's for?galsapir.github.io
1galsapir7mo1 comment

All submissions and comments on Hacker News.