List your company

ResearchFormerly WizWand

Instruction Following

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

Alpaca-Eval (test)

66.85Length-Controlled Winrate

6

Feb 26, 2026

HaluEval QAmis (test)

Three-Stage Fine-Tuning Method

0.008Failure Rate

6

Feb 26, 2026

HaluEval (test)

Three-Stage Fine-Tuning Method

0.36Failure Rate (Sum)

6

Feb 26, 2026

MT-bench and AlpacaEval

1.55Aggregated P

6

Feb 26, 2026

INSTRUCTEVAL (test)

47.3MMLU

6

Feb 26, 2026

Fed-WildChat Real Dataset 1.0 (test)

4.78MT-Bench Score

6

Feb 26, 2026

AlpacaEval Leaderboard Mixtral 8x7B vs Mixtral 8x22B reference (test)

TRLM-FoBa (Rev)

44.28LC Win Rate

6

Feb 26, 2026

85.48Accuracy

6

Feb 26, 2026

Instruction Following SFT 1.0 (eval)

Random Sampling KD (Ours 12)

59.4SFT Score

6

Feb 26, 2026

56.1GPT-4o Score

5

Jul 2, 2026

33.3Utility Score

5

Jun 9, 2026

IFEval, IFBench, and AdvancedIF (test)

84.8IFEval Score

5

Jun 4, 2026

Minecraft E2 single-task (evaluation)

61.1Single-SR

5

Jun 2, 2026

35.12Strict Accuracy

5

May 21, 2026

IFBench Out-of-Distribution

41.5Initial F1

5

Apr 28, 2026

IFBench In-Distribution

41.1Initial F1

5

Apr 28, 2026

Human Performance

91Average Task Success Rate

5

Apr 23, 2026

T5-LM on Unnatural Instructions + Instruction Paraphrases

49Accuracy

5

Feb 27, 2026

AlpacaEval LC (test)

84.8Win Rate

5

Feb 26, 2026

AlpacaEval (OOD)

GEB-arctanh(π − 1)

24.9KL Div (α=1)

5

Feb 26, 2026

Qwen3-VL Thinking

8.16Score

5

Feb 26, 2026

3.99MT-Bench Score

5

Feb 26, 2026

4.24MT-Bench Score

5

Feb 26, 2026

0.614Score

5

Feb 26, 2026

63.8Instruction Following Score

5

Feb 26, 2026