List your company

ResearchFormerly WizWand

Language Modeling

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

LLaMA-2 Family Evaluation v2 (test)

4.8PPL

10

Feb 26, 2026

Language Modeling (test)

33.19PPL (64)

10

Jul 7, 2026

General Text Generation

7.91Fluency

10

Feb 26, 2026

1.762Overall Loss

10

Feb 26, 2026

Mandarin Tail Entity (test)

36.41WER

10

Feb 26, 2026

Mandarin Speech-to-Text (STT) (test)

5.94WER

10

Feb 26, 2026

Mandarin Voice Assistant (VA) (test)

0.051WER

10

Feb 26, 2026

15.31PPLX

10

Feb 26, 2026

MSC Session Openings 1.0 (val)

SumMem-MSC 2.7B

7.78Perplexity

10

Feb 26, 2026

MSC Session 5 1.0 (val)

SumMem-MSC 2.7B

8.99Perplexity

10

Feb 26, 2026

MSC Session 4 1.0 (val)

SumMem-MSC 2.7B

9.07Perplexity

10

Feb 26, 2026

MSC Session 3 1.0 (val)

SumMem-MSC 2.7B

8.96Perplexity

10

Feb 26, 2026

MSC Session 2 1.0 (val)

SumMem-MSC 2.7B

9.08Perplexity

10

Feb 26, 2026

MSC Session 1 1.0 (val)

8.14Perplexity

10

Feb 26, 2026

pushshift.io Reddit (test)

Deeper Transformer (more compute)

22.78Perplexity

10

Feb 26, 2026

pushshift.io Reddit (val)

Deeper Transformer (more compute)

22.72PPL

10

Feb 26, 2026

One Billion Word Benchmark

BIGLSTM baseline

35.1Perplexity

10

Feb 26, 2026

4.74Perplexity

9

Jul 9, 2026

10 LLM benchmark suite macro-average

71.8PASS-AT-1 Accuracy

9

Jun 17, 2026

OpenWebText 1B tokens

26.67Cumulative Orthogonalization Cost (PFLOPs)

9

Jun 16, 2026

15.42Perplexity (C4)

9

Jun 11, 2026

GPT-2 Medium (val)

87.89Runtime

9

Jun 9, 2026

OpenWebText Regime WR (val)

52Time to Target (min)

9

Jun 4, 2026

6.32Perplexity (PPL)

9

May 29, 2026

8.53Perplexity

9

May 29, 2026