INDEX

Explanations

actions, commands, or directives related to decision-making and activity

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

promoting simplicity

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

 estekak

-0.46

rinfo

-0.46

centralwidget

-0.44

ต่างๆ

-0.43

 utafitiHapana

-0.42

URLException

-0.42

 للمعارف

-0.40

 <>",

-0.40

exitRule

-0.39

Istorija

-0.39

POSITIVE LOGITS

 ویکی‌پدیا

0.59

#![

0.48

awtextra

0.47

Personendaten

0.45

ftagPool

0.43

!).

0.42

 Chwiliwch

0.42

kháu

0.41

jboss

0.40

Activations Density 0.049%

actions, commands, or directives related to decision-making and activity

promoting simplicity

No Comments

No Known Activations