INDEX

Explanations

terms expressing high levels of intensity or emphasis

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 1-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_1/width_16k/average_l0_69

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.1.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

Hentet

-0.47

postsleuth

-0.46

 creș

-0.43

makeText

-0.43

 مشين

-0.40

twimg

-0.38

borderSide

-0.38

 TestBed

-0.35

 Guild

-0.35

simpleType

-0.35

POSITIVE LOGITS

BibitemShut

0.61

warzys

0.57

Tet

0.55

 Receipts

0.53

hilangan

0.53

ously

0.52

consciousness

0.50

 upstairs

0.50

uba

0.50

 receipts

0.50

Activations Density 0.408%

terms expressing high levels of intensity or emphasis

No Comments

No Known Activations

terms expressing high levels of intensity or emphasis

No Comments

No Known Activations