Browse › Natural Language Processing › Common Sense Reasoning › ARC (Easy)

Common Sense Reasoning archive 2025-07-28

ARC (Easy) Benchmark (Common Sense Reasoning)

47 rows 41 with code listed 1 metric Dataset page

Common sense reasoning tasks are intended to require the model to go beyond pattern recognition. Instead, the model should use "common sense" or world knowledge to make inferences.

The archive carries no text for this table; the description above is the archive's text for the task Common Sense Reasoning. archive 2025-07-28

Over time archive 2025-07-28

The chart needs JavaScript; the table below carries every value.

Direction inferred from the metric name, not from the archive: Accuracy (higher is better). Points are placed at the row's paper date; 47 of 47 rows carry one.

Results archive 2025-07-28

Archive rows end at the archive snapshot, 2025-07-28: no result published after that date is in this table. Rank is the archive's row order at that snapshot; not re-ranked here. Metric values are the archive's strings. Column headers sort the table in your browser; each row keeps its archive rank.

Paper Code Ran Syntology Report
1 ST-MoE-32B 269B (fine-tuned) 95.2 – Paper Code 2022 5 of 5 ran · 0 unverified report
2 LLaMA 3 8B+MoSLoRA (fine-tuned) 90.5 – Paper Code 2024 4 of 6 ran · 2 unverified report
3 PaLM 2-L (1-shot) 89.7 – Paper Code 2023 linked, not harvested report
4 PaLM 2-M (1-shot) 88.0 – Paper Code 2023 linked, not harvested report
5 LLaMA-3 8B + MixLoRA 86.5 – Paper Code 2024 6 of 11 ran · 5 unverified report
6 Camelidae-8×34B 86.2 – Paper Code 2024 linked, not harvested report
7 PaLM 2-S (1-shot) 85.6 – Paper Code 2023 linked, not harvested report
8 LLaMA 65B + CFG (0-shot) 84.2 – Paper – 2023 no code linked report
9 GAL 120B (0-shot) 83.8 – Paper Code 2022 0 of 2 ran · 2 unverified report
10 LLaMA-2 13B + MixLoRA 83.5 – Paper Code 2024 6 of 11 ran · 5 unverified report
11 LLaMA 30B + CFG (0-shot) 83.2 – Paper – 2023 no code linked report
12 Mixtral 8x7B (0-shot) 83.1 – Paper Code 2024 5 of 5 ran · 0 unverified report
13 FLAN 137B (few-shot, k=14) 80.7 – Paper Code 2021 0 of 1 ran · 1 unverified report
14 Mistral 7B (0-shot) 80.5 – Paper Code 2024 5 of 5 ran · 0 unverified report
15 LLaMA 33B (0-shot) 80.0 – Paper Code 2023 26 of 58 ran · 32 unverified report
16 Mistral 7B (0-shot) 80.0 – Paper Code 2023 9 of 11 ran · 2 unverified report
17 FLAN 137B (0-shot) 79.6 – Paper Code 2021 0 of 1 ran · 1 unverified report
18 LLaMA 13B + CFG (0-shot) 79.1 – Paper – 2023 no code linked report
19 LLaMA 65B (0-shot) 78.9 – Paper Code 2023 26 of 58 ran · 32 unverified report
20 LLaMA-2 7B + MixLoRA 77.7 – Paper Code 2024 6 of 11 ran · 5 unverified report
21 phi-1.5-web 1.3B (0-shot) 76.1 – Paper Code 2023 linked, not harvested report
22 BLOOM 176B (1-shot) 75.93 – Paper Code 2023 linked, not harvested report
23 ST-MoE-L 4.1B (fine-tuned) 75.4 – Paper Code 2022 5 of 5 ran · 0 unverified report
24 GLaM (64B/64E) (5-shot) 74.8 – Paper – 2021 no code linked report
25 LLaMA 13B (0-shot) 74.8 – Paper Code 2023 26 of 58 ran · 32 unverified report
26 Bloomberg GPT 50B (1-shot) 73.99 – Paper Code 2023 linked, not harvested report
27 LLaMA 7B (0-shot) 72.8 – Paper Code 2023 26 of 58 ran · 32 unverified report
28 Pythia 12B (5-shot) 71.5 – Paper Code 2023 linked, not harvested report
29 OPT 66B (1-shot) 71.25 – Paper Code 2023 linked, not harvested report
30 GPT-3 175B (1 shot) 71.2 – Paper Code 2020 15 of 65 ran · 50 unverified report
31 OPT-175B 71.04 – Paper Code 2023 2 of 12 ran · 10 unverified report
32 GPT-NeoX 20B (1-shot) 70.79 – Paper Code 2023 linked, not harvested report
33 Pythia 12B (0-shot) 70.2 – Paper Code 2023 linked, not harvested report
34 UL2 20B (chain-of-thought + self-consistency) 69.8 – Paper Code 2022 0 of 16 ran · 16 unverified report
35 Mamba-2.8B (0-shot) 69.7 – Paper Code 2023 18 of 62 ran · 44 unverified report
36 SparseGPT 175B (50% sparsity) 69.65 – Paper Code 2023 2 of 12 ran · 10 unverified report
37 GPT-3 (zero-shot) 68.8 – Paper Code 2022 0 of 2 ran · 2 unverified report
38 GPT-3 175B (0-shot) 68.8 – Paper Code 2020 15 of 65 ran · 50 unverified report
39 SparseGPT (175B, 4:8 Sparsity) 68.35 – Paper Code 2023 2 of 12 ran · 10 unverified report
40 GLaM 64B/64E (0-shot) 68.0 – Paper – 2021 no code linked report
41 SparseGPT 175B (2:4 sparsity) 67.08 – Paper Code 2023 2 of 12 ran · 10 unverified report
42 LLaMA 7B + CFG (0-shot) 58.9 – Paper – 2023 no code linked report
43 BLOOM (5-shot) 40.7 – Paper Code 2022 0 of 2 ran · 2 unverified report
44 UL2 20B (chain-of-thought) 38.4 – Paper Code 2022 0 of 16 ran · 16 unverified report
45 OPT (5-shot) 37.4 – Paper Code 2022 0 of 2 ran · 2 unverified report
46 UL2 20B (0-shot) 32.2 – Paper Code 2022 0 of 16 ran · 16 unverified report
47 OPT 175B (50% Sparsity) 28.03 – Paper Code 2023 2 of 12 ran · 10 unverified report

All 47 rows shown. 47 link to a paper page on this site; 0 are marked as using additional training data in the archive. No GitHub stars are tracked; "Code" is the first repository the archive lists for the row. The archive carries no row tags, review links or community-submitted rows for this table; none are shown. archive 2025-07-28

Syntology Ran reads "N of M ran · U unverified": of the M code samples Syntology harvested from repositories linked to that row's paper (joined by arXiv id), N executed on a synthesized input and the other U = M−N are unverified (harvested, no recorded run). It counts code from repositories linked to that row's paper, not this result: the row's number was not reproduced and nothing here is a correctness claim. The other cell texts mean no graph line for the row: "linked, not harvested" (the archive links code, Syntology has not harvested it), "no code linked" (no code link in the archive), "not matched" (the row's paper URL matched no paper on this site). 30 rows have a graph line, from 12 distinct papers; 21 rows (9 papers) have at least one sample that ran. Counting each paper once: Syntology ran 90 of 254 samples; 164 unverified. Separately, 57 of those 254 are pointer-only (licence): the site points at that code rather than redistributing it, a licence property recorded for ran and unverified samples alike; each cell's tooltip carries the row's own pointer-only count. Read from the graph 2026-09-24. Per-sample status is on the paper page.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections