HypotheSAEs
hypothesaes.org · 2,476 words · saved by 2 readers
Click to see neuron details!
280-0.189mentions acts of kindness, community improvement, or inspiring collaboration0.750 265-0.158asks the reader a direct question0.990 234-0.140is phrased as a question0.969 272-0.114mentions climate change, global warming, deforestation, or environmental destruction0.734 277-0.112explains or critiques societal systems or problems and suggests the need for change0.741 141-0.099explicitly advocates for an increase or improvement of something0.760 196-0.093uses uplifting or inspirational language to convey a positive or empowering message0.700 106-0.093explicitly focuses on helping…
saved by
related reading
- Hypertext as Literaturehypertext.joodaloop.com
- Anthropic/values-in-the-wild · Datasets at Hugging Facehuggingface.co
- What 81,000 people want from AI \ Anthropicanthropic.com
- Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnettransformer-circuits.pub
- Hacker Newsnews.ycombinator.com
- the singularity is nearer | A home for poorly researched ideas that I find myself repeating a lot anywaygeohot.github.io
- Quantifying Truesight With SAEs · Gwern.netgwern.net
- neuroglancerh01-dot-neuroglancer-demo.appspot.com
- Folder: thoughtsjzhao.xyz
- TextFXtextfx.withgoogle.com
- models have some pretty funny attractor states — LessWronglesswrong.com
- Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnettransformer-circuits.pub