INDEX

Explanations

references to individuals or people 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 @ 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

 Martel

-0.42

 Brew

-0.41

 pregi

-0.41

 gebre

-0.41

Kết

-0.40

oblotting

-0.39

 initial

-0.39

ContentLoaded

-0.38

SharedCtor

-0.38

mix

-0.38

POSITIVE LOGITS

 perſon

1.17

 person

1.09

Person

1.02

 Person

0.97

 PERSON

0.96

person

0.96

people

0.94

 someone

0.92

 persons

0.91

someone

0.90

Activations Density 0.073%

references to individuals or people in various contexts

No Comments

No Known Activations