INDEX

Explanations

specific references or prompts that indicate user engagement or interaction

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

user input following turn start

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

thâu

-0.53

ыгана

-0.52

opatra

-0.51

ouncements

-0.49

 beft

-0.49

 مشين

-0.48

LabelTagHelper

-0.48

Olig

-0.48

dflare

-0.47

potranspiration

-0.46

POSITIVE LOGITS

 Farbe

0.36

setVerticalGroup

0.34

 escrita

0.33

 stylized

0.32

 Verarbeitung

0.31

 written

0.31

utilisons

0.31

 instalada

0.31

 Schuhe

0.31

 slanted

0.31

Activations Density 0.000%

specific references or prompts that indicate user engagement or interaction

user input following turn start

No Comments

No Known Activations