INDEX

Explanations

instances of the word "to" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

google/gemma-scope-9b-pt-res/layer_2/width_16k/average_l0_67

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.2.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

นะครับ

-0.51

 alimentaria

-0.50

 débil

-0.47

retudo

-0.46

 uppgifter

-0.44

 satisfecho

-0.44

 interesantes

-0.43

 democrática

-0.42

 maravillosa

-0.42

 merkezi

-0.42

POSITIVE LOGITS

zA

0.82

casio

0.68

non

0.68

non

0.65

than

0.62

 dintr

0.60

 self

0.60

 faſt

0.59

mid

0.59

 semi

0.58

Activations Density 0.195%

instances of the word "to" in various contexts

No Comments

No Known Activations