INDEX

Explanations

compound words or specific terms that describe concepts or categories related to various subjects

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

hspace

-0.52

ът

-0.51

(“

-0.50

fahan

-0.49

alış

-0.48

«<

-0.48

woff

-0.47

multicolumn

-0.47

ฒ

-0.46

frac

-0.46

POSITIVE LOGITS

ey

0.93

0.86

ry

0.85

yyyy

0.85

ety

0.83

ty

0.79

yyy

0.78

edy

0.76

tty

0.76

tomy

0.74

Activations Density 0.994%

compound words or specific terms that describe concepts or categories related to various subjects

No Comments

No Known Activations