INDEX

Explanations

phrases indicating partial approval or denial of requests

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

granted in part and denied in part

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

Rigidbody

-0.37

DockStyle

-0.36

yelidikan

-0.35

 referrerpolicy

-0.34

ancang

-0.34

 leña

-0.33

 violencia

-0.32

 hẳn

-0.32

 vitales

-0.32

ilaian

-0.32

POSITIVE LOGITS

setVerticalGroup

0.67

parsedMessage

0.64

 ModelExpression

0.58

Personendaten

0.56

rungsseite

0.55

 betweenstory

0.55

esModule

0.53

ChildScrollView

0.52

PerformLayout

0.51

帖最后由

0.50

Activations Density 0.034%

phrases indicating partial approval or denial of requests

granted in part and denied in part

No Comments

No Known Activations