INDEX

Explanations

punctuated phrases that indicate dialogue or speech

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

words like ok, honey, are

np_acts-logits-general · gemini-2.5-flash-lite

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 autorytatywna

-0.63

adaptiveStyles

-0.56

 pinulongan

-0.51

 MainAxisSize

-0.50

StoryboardSegue

-0.49

PerformLayout

-0.48

Enllaces

-0.47

imágenes

-0.45

 betweenstory

-0.45

vább

-0.44

POSITIVE LOGITS

GEBURTS

0.52

if

0.50

testify

0.42

gnac

0.41

 verhe

0.41

Dere

0.40

olkien

0.39

oy

0.39

anın

0.39

ukone

0.39

Activations Density 0.008%

punctuated phrases that indicate dialogue or speech

words like ok, honey, are

No Comments

No Known Activations