Benchmark Almanac

Mikhail Doroshenko

We track 12 benchmarks that Mikhail Doroshenko made or helped make.

0 in use 0 with public code 12 with a leaderboard here

Benchmarks

Name Kind Released Best score Status
The Aggregate Day-one Recall Check Models answer 34 questions about 12 public-domain books from memory. The scorer matches answers against a key and excludes empty replies. Trivia and open questions Aug 2026 61.8% Not in cards
The Aggregate Trip Outcome Prediction Models read a packing list and itinerary, then guess what happened on a weekend trip. A judge scores four answers, and relative error scores the taxi fare. Commonsense Nov 2026 69.8% Not in cards
The Aggregate Puzzle Portfolio Ladder Models solve 30 generated puzzles across five families. Each family has six questions with answers fixed by the generator. Puzzles and chess 96.7% Not in cards
Guesswork (The Aggregate) Models predict benchmark scores before publication. The maker ranks forecasts by mean absolute error after scaling each error by the variation in that benchmark’s scores. Real-world events Jun 2026 0.56 Not in cards
The Aggregate Long-tail Book Recall Models answer closed-book questions about 60 Project Gutenberg books. The scorer checks the answer text, and a second reviewer can accept equivalent wording. Trivia and open questions 77% Not in cards
The Aggregate Inversion and Execution Ladder Models predict the output of a short program or reconstruct the input to a digit transformation. The scorer checks exact answers across six difficulty levels. Puzzles and chess 100% Not in cards
The Aggregate Lichess Pairwise Strength Models read two anonymized chess games and choose which designated player had the higher pregame Lichess rapid rating. Accuracy counts 25 choices. Puzzles and chess 88% Not in cards
The Aggregate Lichess Rating Prediction Models estimate both players' pregame Lichess rapid ratings from the moves of 25 anonymized games. The score is mean absolute error across 50 players. Puzzles and chess 270 Not in cards
The Aggregate Personal Style Imitation Models rewrite 53 neutral drafts in one person's style using 30 other messages as examples. An authorship classifier scores each rewrite. Creative writing Nov 2026 76.7% Not in cards
The Aggregate Exotic Fact Recall Models recall exact fields from five public registries without looking them up. The bank mixes familiar entries with randomly sampled obscure ones. Trivia and open questions Sep 2026 75% Not in cards
The Aggregate Asteroid Numbers Models give the permanent numbers of 48 named asteroids from memory. The scorer checks each answer against the NASA Jet Propulsion Laboratory catalog. Trivia and open questions Sep 2026 58.3% Not in cards
The Aggregate Forecasting Resolved Questions Models estimate probabilities for resolved Manifold markets. The author drops questions when a model correctly recalls the outcome, then scores the remaining forecasts by squared error. Real-world events Sep 2026 0.27 Not in cards