viemccoy

Karma: 64

The Weighted Perplexity Benchmark: Tokenizer-Normalized Evaluation for Language Model Comparison

jessicata and viemccoy

7 Jul 2025 21:43 UTC

21 points

0 comments7 min readLW link

(www.morpheus.systems)

Schizobench: Documenting Magical-Thinking Behavior in Claude 4 Opus

viemccoy23 May 2025 1:31 UTC

23 points

0 comments1 min readLW link

(metanomicon.ink)

Defense Against The Super-Worms

viemccoy20 Mar 2025 7:24 UTC

24 points

1 comment2 min readLW link