INDEX

Explanations

phrases that indicate emphasis or importance

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

+#+

-0.52

RTLI

-0.50

 artesanales

-0.46

 plegable

-0.43

{},

-0.41

jenigen

-0.39

 bordada

-0.39

 mijne

-0.38

 obreros

-0.38

 Roskov

-0.38

POSITIVE LOGITS

msgTypes

0.63

 Infórmanos

0.63

 betweenstory

0.59

 estekak

0.58

AutoScale

0.55

SequentialGroup

0.50

verifyException

0.50

SourceChecksum

0.47

riwal

0.46

 Exacts

0.45

Activations Density 0.251%

phrases that indicate emphasis or importance

No Comments

No Known Activations