INDEX

Explanations

features related to functionality and services in various applications

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

features and functions

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

AutoScaleMode

-0.45

InitVars

-0.36

')],

-0.35

 vuitton

-0.33

']>;

-0.30

 hinein

-0.30

labelledby

-0.30

zedł

-0.30

Derbyniad

-0.30

'},

-0.30

POSITIVE LOGITS

+#+#

0.53

 CURLOPT

0.52

confirmButton

0.52

freezing

0.51

reddits

0.50

frontal

0.49

allow

0.48

Spoljašnje

0.48

 loophole

0.48

launcher

0.48

Activations Density 0.035%

features related to functionality and services in various applications

features and functions

No Comments

No Known Activations