INDEX

Explanations

numerical results, statements, and punctuations indicating conclusion or emphasis

oai_token-act-pair · gpt-4o-mini Triggered by @bot

end of sentence then word

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

DoubleQuotes

-0.57

 المعيارى

-0.43

锈钢

-0.42

 للاسماء

-0.41

новништво

-0.41

dataclass

-0.39

AnimationsModule

-0.38

InjectAttribute

-0.38

 kasarigan

-0.37

afficheront

-0.37

POSITIVE LOGITS

 contextLoads

0.52

wieder

0.44

 CGRectMake

0.43

imageshack

0.42

 SPDX

0.42

 бел

0.42

anzi

0.41

lst

0.41

Ueb

0.41

smithy

0.40

Activations Density 0.093%

numerical results, statements, and punctuations indicating conclusion or emphasis

end of sentence then word

No Comments

No Known Activations