Browse › Natural Language Processing › Question Answering › BoolQ
BoolQ Benchmark (Question Answering)
Question answering can be segmented into domain-specific tasks like community question answering and knowledge-base question answering. Popular benchmark datasets for evaluation question answering systems include SQuAD, HotPotQA, bAbI, TriviaQA, WikiQA, and many others. Models for question answering are typically evaluated on metrics like EM and F1. Some recent top performing models are T5 and XLNet.
The archive carries no text for this table; the description above is the archive's text for the task Question Answering. 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; 65 of 65 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 | Mistral-Nemo 12B (HPT) | 99.87 | – | Paper | Code | 2024 | linked, not harvested | report |
| 2 | ST-MoE-32B 269B (fine-tuned) | 92.4 | – | Paper | Code | 2022 | 5 of 5 ran · 0 unverified | report |
| 3 | PaLM 540B (fine-tuned) | 92.2 | – | Paper | Code | 2022 | 32 of 37 ran · 5 unverified | report |
| 4 | Turing NLR v5 XXL 5.4B (fine-tuned) | 92 | – | Paper | – | 2022 | no code linked | report |
| 5 | T5-XXL 11B (fine-tuned) | 91.2 | – | Paper | Code | 2019 | 21 of 31 ran · 10 unverified | report |
| 6 | PaLM 2-L (1-shot) | 90.9 | – | Paper | Code | 2023 | linked, not harvested | report |
| 7 | UL2 20B (fine-tuned) | 90.8 | – | Paper | Code | 2022 | 15 of 16 ran · 1 unverified | report |
| 8 | Vega v2 6B (fine-tuned) | 90.5 | – | Paper | – | 2022 | no code linked | report |
| 9 | DeBERTa-1.5B | 90.4 | – | Paper | Code | 2020 | 4 of 13 ran · 9 unverified | report |
| 10 | PaLM 2-M (1-shot) | 88.6 | – | Paper | Code | 2023 | linked, not harvested | report |
| 11 | ST-MoE-L 4.1B (fine-tuned) | 88.6 | – | Paper | Code | 2022 | 5 of 5 ran · 0 unverified | report |
| 12 | PaLM 2-S (1-shot) | 88.1 | – | Paper | Code | 2023 | linked, not harvested | report |
| 13 | MUPPET Roberta Large | 87.5 | – | Paper | Code | 2021 | linked, not harvested | report |
| 14 | FLAN 137B (prompt-tuned) | 86.3 | – | Paper | Code | 2021 | 0 of 1 ran · 1 unverified | report |
| 15 | RoBERTa-large 355M + Entailment as Few-shot Learner | 86.0 | – | Paper | Code | 2021 | 1 of 3 ran · 2 unverified | report |
| 16 | T5-Large 770M (fine-tuned) | 85.4 | – | Paper | Code | 2019 | 21 of 31 ran · 10 unverified | report |
| 17 | LLaMA 65B (0-shot) | 85.3 | – | Paper | Code | 2023 | 37 of 58 ran · 21 unverified | report |
| 18 | LLaMA 2 70B (0-shot) | 85 | – | Paper | Code | 2023 | 31 of 52 ran · 21 unverified | report |
| 19 | FLAN 137B (4-shot) | 84.6 | – | Paper | Code | 2021 | 0 of 1 ran · 1 unverified | report |
| 20 | MUPPET Roberta Base | 83.8 | – | Paper | Code | 2021 | linked, not harvested | report |
| 21 | Chinchilla 70B (0-shot) | 83.7 | – | Paper | Code | 2022 | 8 of 11 ran · 3 unverified | report |
| 22 | LLaMA 2 34B (0-shot) | 83.7 | – | Paper | Code | 2023 | 31 of 52 ran · 21 unverified | report |
| 23 | LLaMA 33B (0-shot) | 83.1 | – | Paper | Code | 2023 | 37 of 58 ran · 21 unverified | report |
| 24 | FLAN 137B (0-shot) | 82.9 | – | Paper | Code | 2021 | 0 of 1 ran · 1 unverified | report |
| 25 | LLaMA 2 13B (0-shot) | 81.7 | – | Paper | Code | 2023 | 31 of 52 ran · 21 unverified | report |
| 26 | T5-Base 220M (fine-tuned) | 81.4 | – | Paper | Code | 2019 | 21 of 31 ran · 10 unverified | report |
| 27 | BERT-MultiNLI 340M (fine-tuned) | 80.4 | – | Paper | Code | 2019 | linked, not harvested | report |
| 28 | Gopher (zero-shot) | 79.3 | – | Paper | Code | 2021 | linked, not harvested | report |
| 29 | LLaMA 13B (zero-shot) | 78.1 | – | Paper | Code | 2023 | 37 of 58 ran · 21 unverified | report |
| 30 | LLaMA 2 7B (zero-shot) | 77.4 | – | Paper | Code | 2023 | 31 of 52 ran · 21 unverified | report |
| 31 | LLaMA-2 13B + MixLoRA | 77.1 | – | Paper | Code | 2024 | 6 of 11 ran · 5 unverified | report |
| 32 | LLaMA 7B (zero-shot) | 76.5 | – | Paper | Code | 2023 | 37 of 58 ran · 21 unverified | report |
| 33 | T5-Small 60M (fine-tuned) | 76.4 | – | Paper | Code | 2019 | 21 of 31 ran · 10 unverified | report |
| 34 | GPT-3 175B (few-shot, k=32) | 76.4 | – | Paper | Code | 2020 | 41 of 65 ran · 24 unverified | report |
| 35 | BiDAF-MultiNLI (fine-tuned) | 75.57 | – | Paper | Code | 2019 | linked, not harvested | report |
| 36 | LLaMA-3 8B + MixLoRA | 75 | – | Paper | Code | 2024 | 6 of 11 ran · 5 unverified | report |
| 37 | Bloomberg GPT 50B (1-shot) | 74.6 | – | Paper | Code | 2023 | linked, not harvested | report |
| 38 | LLaMA3+MoSLoRA | 74.6 | – | Paper | Code | 2024 | 4 of 6 ran · 2 unverified | report |
| 39 | GPT-1 117M (fine-tuned) | 72.87 | – | Paper | Code | 2019 | linked, not harvested | report |
| 40 | LLaMA-2 7B + MixLoRA | 72.7 | – | Paper | Code | 2024 | 6 of 11 ran · 5 unverified | report |
| 41 | BiDAF + ELMo (fine-tuned) | 71.41 | – | Paper | Code | 2019 | linked, not harvested | report |
| 42 | OPT-IML 175B | 71.4 | – | Paper | Code | 2022 | linked, not harvested | report |
| 43 | AlexaTM 20B | 69.4 | – | Paper | Code | 2022 | 1 of 1 ran · 0 unverified | report |
| 44 | Neo-6B (QA + WS) | 67.2 | – | Paper | Code | 2022 | 2 of 2 ran · 0 unverified | report |
| 45 | OPT-IML 30B | 66.9 | – | Paper | Code | 2022 | linked, not harvested | report |
| 46 | Neo-6B (few-shot) | 66.5 | – | Paper | Code | 2022 | 2 of 2 ran · 0 unverified | report |
| 47 | N-Grammer 343M | 65 | – | Paper | Code | 2022 | 6 of 6 ran · 0 unverified | report |
| 48 | Neo-6B (QA) | 64.9 | – | Paper | Code | 2022 | 2 of 2 ran · 0 unverified | report |
| 49 | OPT 30B (0-shot) | 64 | – | Paper | Code | 2022 | linked, not harvested | report |
| 50 | UL2 20B (0-shot) | 63.1 | – | Paper | Code | 2022 | 15 of 16 ran · 1 unverified | report |
| 51 | Majority baseline | 62.17 | – | Paper | Code | 2019 | linked, not harvested | report |
| 52 | Hybrid H3 1.3B (0-shot, logit scoring) | 61.7 | – | Paper | Code | 2022 | 7 of 15 ran · 8 unverified | report |
| 53 | OPT-IML 1.3B (0-shot) | 61.5 | – | Paper | Code | 2022 | linked, not harvested | report |
| 54 | Shakti-LLM (2.5B) | 61.1 | – | Paper | – | 2024 | no code linked | report |
| 55 | Hybrid H3 2.7B (3-shot, logit scoring) | 60.6 | – | Paper | Code | 2022 | 7 of 15 ran · 8 unverified | report |
| 56 | OPT 1.3B (zero-shot) | 60.5 | – | Paper | Code | 2022 | linked, not harvested | report |
| 57 | GPT-3 75B (0-shot) | 60.5 | – | Paper | Code | 2020 | 41 of 65 ran · 24 unverified | report |
| 58 | OPT 175B | 60.1 | – | Paper | Code | 2022 | linked, not harvested | report |
| 59 | Hybrid H3 125M (0-shot, logit scoring) | 59.6 | – | Paper | Code | 2022 | 7 of 15 ran · 8 unverified | report |
| 60 | OPT 66B (1-shot) | 57.5 | – | Paper | Code | 2023 | linked, not harvested | report |
| 61 | Hybrid H3 125M (3-shot, logit scoring) | 56.1 | – | Paper | Code | 2022 | 7 of 15 ran · 8 unverified | report |
| 62 | Hybrid H3 125M (3-shot, rank classification) | 56.1 | – | Paper | Code | 2022 | 7 of 15 ran · 8 unverified | report |
| 63 | BLOOM 176B (1-shot) | 52.9 | – | Paper | Code | 2023 | linked, not harvested | report |
| 64 | Hyena | 51.8 | – | Paper | Code | 2023 | 5 of 5 ran · 0 unverified | report |
| 65 | GPT-NeoX 20B (1-shot) | 46.4 | – | Paper | Code | 2023 | linked, not harvested | report |
All 65 rows shown. 65 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). 40 rows have a graph line, from 18 distinct papers; 37 rows (17 papers) have at least one sample that ran. Counting each paper once: Syntology ran 226 of 338 samples; 112 unverified. Separately, 46 of those 338 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 2026-09-25. 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