INDEX

Explanations

phrases referencing the concept of being "inside" or "interior"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

 alumínio

-0.50

 preghiera

-0.48

 frango

-0.48

 dårlig

-0.47

 femminile

-0.46

差不

-0.46

RefNanny

-0.45

 africains

-0.44

Liam

-0.44

 debole

-0.44

POSITIVE LOGITS

inside

1.00

Inside

0.98

 INSIDE

0.96

 Inside

0.96

 inside

0.95

INSIDE

0.88

 OUTSIDE

0.81

outside

0.77

 Outside

0.73

Outside

0.71

Activations Density 0.066%

phrases referencing the concept of being "inside" or "interior"

No Comments

No Known Activations