INDEX

Explanations

quantitative data related to measurements and numerical values

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

numbers and types

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

-0.46

-0.35

 points

-0.34

 nomb

-0.33

 arkas

-0.33

it

-0.33

ลง

-0.32

 structure

-0.32

-0.31

Do

-0.31

POSITIVE LOGITS

featureID

0.82

$_"

0.81

 Administrativna

0.79

 queſta

0.76

iſchen

0.72

LEncoder

0.71

ImageContext

0.70

 beſch

0.68

ロウィン

0.67

 Wikimedijinoj

0.67

Activations Density 0.207%

quantitative data related to measurements and numerical values

numbers and types

No Comments

No Known Activations