INDEX

Explanations

references to specific brands, names, or entities

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

proper names

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

 pré

-0.49

hitheatre

-0.48

__':

-0.47

Rohy

-0.47

cot

-0.46

gdx

-0.45

 sveta

-0.44

 ageing

-0.44

ge

-0.43

فایل‌لار

-0.43

POSITIVE LOGITS

Verizon

1.25

 Verizon

1.20

 Verſ

0.68

 verſ

0.66

 ainfi

0.60

izon

0.58

 plufieurs

0.58

ientras

0.56

 Monfieur

0.55

 vectorielles

0.54

Activations Density 0.000%

references to specific brands, names, or entities

proper names

No Comments

No Known Activations