Gpt2 Xl — Benchmarks

Benchmark scores for Gpt2 Xl aggregated from public leaderboards (7 boards), with how it ranks among open models and how it compares with its nearest neighbours. See hardware requirements for what you need to run it.

No llmrun Composite Score for Gpt2 Xl: the score is only published for a model measured on at least 4 of the boards in the score panel, spanning at least 2 benchmark categories and including a reasoning or coding board — or on 3 boards that narrow its score to a 90% range of at most 15.6 points. Either way the results must come from at least two different benchmarks. Its individual scores below stand on their own — see how the llmrun Composite Score gets calculated.

Gpt2 Xl reasoning benchmarks

Multi-step logic and problem-solving — can the model think through a hard problem rather than recall an answer.

WinoGrande
WinoGrande is a large-scale commonsense-reasoning test built around pronoun resolution: the model must pick which of two nouns a pronoun refers to in a sentence, using real-world knowledge to disambiguate. It was designed by AI2 to resist the shortcuts that let earlier models game its predecessor, Winograd.
58.3%
58.3%
median 70.1%open leader 89.2%
14th percentile · #44 of 51 open · #72 of 79 all
#44 of 51
#72 of 79
Lab-reported
Epoch AI · Nov 2019

Other results (1)· leaderboard-only, not part of the llmrun Score

PIQA
PIQA (Physical Interaction QA) asks a model to choose the more sensible of two ways to accomplish an everyday physical task, testing practical, physical-world commonsense that isn't well captured by purely textual reasoning tests.
70.5%
70.5%
median 80.1%open leader 88.6%
Overall best: GPT 4o Mini (Jul 18, 2024) 88.7%
5th percentile · #38 of 40 open · #57 of 59 all
#38 of 40
#57 of 59
Lab-reported
Epoch AI · Nov 2019

Gpt2 Xl coding benchmarks

Writing, editing, and fixing real code — measured by whether the resulting program actually runs and passes tests.

METR Time Horizons
METR Time Horizons, from the AI-evaluation nonprofit METR, runs models on a suite of software-engineering and research tasks whose length is measured by how long they take skilled humans, from seconds to many hours. METR summarises each model as a time horizon: the human task length at which the model succeeds half the time. llmrun shows METR's average task-success score (0–1) as published by Epoch AI; higher means the model completes more of the suite, including longer tasks.
10.1%
10.1%
median 49.6%open leader 59.2%
Overall best: Claude Mythos Preview Early 85.2%
lowest · #9 of 9 open · #46 of 46 all
#9 of 9
#46 of 46
Lab-reported
Epoch AI · Nov 2019

Gpt2 Xl knowledge benchmarks

Breadth and depth of factual knowledge across many subjects, paired with the reasoning to apply it.

ARC Challenge
ARC Challenge is a set of grade-school science questions, curated by the Allen Institute for AI (AI2) to exclude ones that simple word-matching or retrieval can answer. It measures science knowledge paired with the reasoning needed to apply it.
25.0%
25.0%
median 53.9%open leader 95.3%
lowest · #56 of 56 open · #76 of 76 all
#56 of 56
#76 of 76
Lab-reported
Epoch AI · Nov 2019

Other results (3)· leaderboard-only, not part of the llmrun Score

HellaSwag
HellaSwag is a commonsense benchmark that asks a model to pick the most plausible continuation of an everyday situation. The wrong options are adversarially chosen to fool models, testing grounded commonsense understanding.
40.0%
40.0%
median 76.2%open leader 89.2%
Overall best: GPT 4 (Mar 14) 95.3%
lowest · #47 of 47 open · #75 of 75 all
#47 of 47
#75 of 75
Lab-reported
Epoch AI · Nov 2019
BoolQ
BoolQ is a reading-comprehension benchmark of naturally occurring yes/no questions, each paired with a short passage that contains the answer. It tests whether a model can extract and reason over information from real text rather than just recall facts.
61.8%
61.8%
median 76.4%open leader 89.4%
Overall best: T5 11B 91.2%
7th percentile · #36 of 38 open · #69 of 76 all
#36 of 38
#69 of 76
Lab-reported
Epoch AI · Nov 2019
OpenBookQA
OpenBookQA is a science-question benchmark modeled on open-book exams: the model is given a small set of core science facts and must combine them with broader common knowledge to answer multi-step questions. It targets reasoning over recall.
22.4%
22.4%
median 56.8%open leader 88.0%
lowest · #26 of 26 open · #41 of 41 all
#26 of 26
#41 of 41
Lab-reported
Epoch AI · Nov 2019

Scores aggregated from public benchmark sources (Epoch AI: 7 boards) — each linked from the benchmark pages. llmrun does not run these benchmarks. Percentiles and ranks are among open models in the llmrun catalogue; the median and leader marks refer to the same field.