Granite 3.3 8B Base

Granite-3.3-8B-Base is a decoder-only language model with a 128K token context window. It improves upon Granite-3.1-8B-Base by adding support for Fill-in-the-Middle (FIM) using specialized tokens, enabling the model to generate content conditioned on both prefix and suffix.

HumanEval

89.7%

i
AttaQ

88.5%

i
HumanEval+

86.1%

i
AIME 2024

81.2%

i
HellaSwag

80.1%

i
TriviaQA

78.2%

i
IFEval

74.8%

i
Winogrande

74.4%

i
BIG-Bench Hard

69.1%

i
MATH-500

69.0%

i
MMLU

63.9%

i
AlpacaEval 2.0

62.7%

i
GSM8k

59.0%

i
Arena Hard

57.6%

i
TruthfulQA

52.1%

i
ARC-C

50.8%

i
AGIEval

49.3%

i
NQ

36.5%

i
DROP

36.1%

i
PopQA

26.2%

i