INDEX

Explanations

phrases related to legal and procedural terms

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

OMI

-0.38

One

-0.38

IMI

-0.37

newcommand

-0.37

Creative

-0.36

lization

-0.36

tilization

-0.36

 nagu

-0.36

ABB

-0.35

POSITIVE LOGITS

RenderAtEndOf

0.70

TypedDataSet

0.65

protoimpl

0.65

AndEndTag

0.63

ArrowToggle

0.62

 betweenstory

0.60

 NSCoder

0.57

ViewImports

0.56

 HasFactory

0.54

HasColumnName

0.53

Activations Density 0.080%

phrases related to legal and procedural terms

No Comments

No Known Activations