INDEX

Explanations

specific patterns related to words with the segment "aw" and names or titles involving "W" and "P"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 21-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_21/width_16k/average_l0_129

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.21.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

 Leger

-0.62

TemporalType

-0.59

retin

-0.59

basicConfig

-0.57

PTI

-0.56

richlet

-0.55

Berger

-0.55

 Lest

-0.55

Argentina

-0.54

 Yeti

-0.54

POSITIVE LOGITS

aw

0.77

ow

0.75

Lw

0.75

ew

0.74

Jow

0.72

Tow

0.66

ww

0.65

Caw

0.65

tow

0.64

Gow

0.64

Activations Density 0.532%

specific patterns related to words with the segment "aw" and names or titles involving "W" and "P"

No Comments

No Known Activations