INDEX

Explanations

references to "address" and related terms in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 5-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_5/width_16k/average_l0_68

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.5.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Fland

-0.70

 Wyman

-0.68

iwa

-0.66

 Campion

-0.66

Tut

-0.66

 Mert

-0.65

Kjelder

-0.65

²/

-0.65

sqcup

-0.63

 vân

-0.63

POSITIVE LOGITS

 address

1.95

 ADDRESS

1.94

 addressed

1.94

 addresses

1.94

 Addresses

1.91

 Address

1.90

address

1.89

Address

1.78

Addresses

1.77

addressed

1.73

Activations Density 0.042%

references to "address" and related terms in various contexts

No Comments

No Known Activations