Gemma 2 27B - NC Bench

google/gemma-2-27b-it

Gemma 2 27B

via OpenRouter

Release Date

Jun 27th, 2024

Parameters

27B

Context Size

Creative writing

23.28%

Rule following

44.79%

Utility

52.72%

Mathematics

95.00%

Tooling

51.28%

Language

80.31%

Logic

80.63%

Data extraction

Extract key details from a given block of text.

Scenario	Run 1	Run 2	Run 3	Run 4	Run 5	Run 6	Run 7	Run 8	Run 9	Run 10	Total
Who's the tallest? 0-shot UtilityLogic	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
What's the color of the car? 0-shot UtilityLogic	100%	100%	100%	100%	100%	100%	100%	100%	100%	50%	95%
What instrument does Lucy play? 0-shot UtilityLogic	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
Guess the pet 0-shot UtilityLogic	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
What's the correct time? 0-shot UtilityLogic	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
Who's the sister? 0-shot UtilityLogic	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
Contextual pronoun 0-shot Utility	100%	100%	100%	100%	100%	100%	100%	100%	100%	0%	90%
Indirect birth year 0-shot UtilityMathematicsLogic	100%	100%	100%	100%	100%	100%	100%	100%	100%	50%	95%
Fruits excluding citrus 0-shot Utility	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
Future event time 0-shot UtilityLogic	100%	50%	50%	50%	50%	50%	50%	50%	50%	50%	55%
Highest-rated movie 0-shot Utility	100%	100%	50%	50%	50%	50%	50%	50%	50%	50%	60%
All valid emails 0-shot Utility	100%	100%	100%	0%	0%	0%	0%	0%	0%	0%	30%
68.75%

Dialogue tags

Various tasks related to dialogue tags in text.

Scenario	Run 1	Run 2	Run 3	Run 4	Run 5	Run 6	Run 7	Run 8	Run 9	Run 10	Total
Write unattributed dialogue 0-shot Creative writingRule following	100%	100%	100%	61%	61%	14%	14%	14%	1%	0%	46%
Write 200 words with 10% dialogue 0-shot Creative writingRule following	86%	50%	50%	50%	48%	0%	0%	0%	0%	0%	28%
Write 200 words with 50% dialogue 0-shot Creative writingRule following	50%	50%	48%	44%	35%	26%	7%	5%	1%	0%	27%
Write 200 words with 90% dialogue 0-shot Creative writingRule following	72%	49%	32%	18%	18%	18%	12%	5%	3%	0%	23%
Write 500 words with 30% dialogue 0-shot Creative writingRule following	52%	50%	0%	0%	0%	0%	0%	0%	0%	0%	10%
Write 500 words with 50% dialogue 0-shot Creative writingRule following	49%	38%	19%	7%	3%	0%	0%	0%	0%	0%	12%
Write 500 words with 70% dialogue 0-shot Creative writingRule following	50%	35%	33%	32%	14%	3%	2%	2%	0%	0%	17%
23.28%

Language Comprehension

Does the model understand more than just English?

Scenario	Run 1	Run 2	Run 3	Run 4	Run 5	Total
Friend got new kittens (Tagalog) 0-shot Language	100%	100%	100%	100%	100%	100%
Friend got new kittens (German) 0-shot Language	100%	100%	100%	100%	100%	100%
Asking for directions (German) 0-shot Language	100%	100%	100%	100%	0%	80%
Asking for directions (Dutch) 0-shot Language	100%	100%	100%	100%	0%	80%
90.00%

Language Writing

Can the model generate text in different languages?

Scenario	Run 1	Run 2	Run 3	Run 4	Run 5	Total
Character dialogue (Spanish) in a story 0-shot Language	83%	82%	64%	62%	54%	69%
Character dialogue (French) in a story 0-shot Language	86%	86%	78%	78%	77%	81%
Character dialogue (German) in a story 0-shot Language	94%	92%	85%	40%	33%	69%
Character dialogue (Italian) in a story 0-shot Language	100%	90%	79%	70%	50%	78%
Character dialogue (Hindi) in a story 0-shot Language	100%	67%	63%	60%	43%	66%
72.56%

Novel outline

Handle questions about the outline of a novel in various formats

Scenario	Run 1	Run 2	Run 3	Run 4	Run 5	Run 6	Run 7	Run 8	Run 9	Run 10	Total
Count chapters 0-shot ToolingUtility	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
Count chapters (indented) 0-shot ToolingUtility	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
Count acts 0-shot ToolingUtility	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
Count acts (indented) 0-shot ToolingUtility	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
Count scenes 0-shot ToolingUtility	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
Count scenes (indented) 0-shot ToolingUtility	100%	100%	100%	0%	0%	0%	0%	0%	0%	0%	30%
Count point of views for Jack Harper 0-shot ToolingUtility	100%	100%	100%	0%	0%	0%	0%	0%	0%	0%	30%
Count point of views for Jack Harper (indented) 0-shot ToolingUtility	100%	100%	100%	100%	100%	0%	0%	0%	0%	0%	50%
Count point of views for Olivia 0-shot ToolingUtility	100%	100%	100%	100%	0%	0%	0%	0%	0%	0%	40%
Count point of views for Olivia (indented) 0-shot ToolingUtility	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
Count point of views for Jack and Olivia 0-shot ToolingUtility	100%	50%	50%	0%	0%	0%	0%	0%	0%	0%	20%
Count point of views for Jack and Olivia (indented) 0-shot ToolingUtility	50%	50%	50%	50%	0%	0%	0%	0%	0%	0%	20%
49.17%

Tool usage within Novelcrafter

Output messages that are related to tool usage within Novelcrafter

Scenario	Run 1	Run 2	Run 3	Run 4	Run 5	Run 6	Run 7	Run 8	Run 9	Run 10	Total
Create alternate prose sections 0-shot ToolingUtility	100%	100%	100%	100%	100%	100%	100%	67%	0%	0%	77%
76.67%

N-Length Sentences

Write sentences with exactly N words

Scenario	Run 1	Run 2	Run 3	Run 4	Run 5	Run 6	Run 7	Run 8	Run 9	Run 10	Total
Write sentences with 5 words each 0-shot Rule following	84%	71%	68%	62%	61%	57%	55%	52%	52%	40%	60%
Write sentences with 10 words each 0-shot Rule following	82%	48%	45%	43%	43%	40%	39%	32%	21%	21%	41%
Write sentences with 20 words each 0-shot Rule following	44%	43%	39%	28%	27%	20%	14%	12%	5%	0%	23%
41.58%

Voice/dialogue sheets

Extract dialogue from given text as voice sheets.

Scenario	Run 1	Run 2	Run 3	Run 4	Run 5	Run 6	Run 7	Run 8	Run 9	Run 10	Total
Simple 0-shot Utility	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
Simple (1-shot) 1-shot Utility	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
Simple (5-shot) Few-shot Utility	100%	100%	100%	100%	100%	100%	100%	100%	100%	0%	90%
Multiple speakers 0-shot Utility	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
Unattributed dialogue 0-shot Utility	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
18.00%

Write N of X

Write exactly N words/sentences/paragraphs...

Scenario	Run 1	Run 2	Run 3	Run 4	Run 5	Run 6	Run 7	Run 8	Run 9	Run 10	Total
10 word summary 0-shot Rule following	100%	100%	100%	100%	100%	100%	98%	98%	98%	98%	99%
20 word summary 0-shot Rule following	100%	100%	100%	100%	100%	98%	92%	54%	54%	54%	85%
50 word summary 0-shot Rule following	98%	54%	0%	0%	0%	0%	0%	0%	0%	0%	15%
100 word summary 0-shot Rule following	100%	0%	0%	0%	0%	0%	0%	0%	0%	0%	10%
200 word summary 0-shot Rule following	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
1 sentence summary 0-shot Rule following	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
3 sentence summary 0-shot Rule following	100%	100%	100%	100%	100%	100%	100%	100%	100%	77%	98%
10 sentence summary 0-shot Rule following	100%	100%	100%	100%	100%	100%	100%	77%	77%	0%	85%
20 sentence summary 0-shot Rule following	100%	100%	100%	100%	98%	98%	92%	0%	0%	0%	69%
50 sentence summary 0-shot Rule following	9%	0%	0%	0%	0%	0%	0%	0%	0%	0%	1%
1 paragraph summary 0-shot Rule following	100%	100%	100%	100%	100%	100%	100%	100%	100%	0%	90%
3 paragraph summary 0-shot Rule following	100%	100%	100%	100%	100%	100%	100%	0%	0%	0%	70%
5 paragraph summary 0-shot Rule following	100%	100%	0%	0%	0%	0%	0%	0%	0%	0%	20%
57.12%