INDEX

Explanations

terms related to addresses in various formats

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 22-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_22/width_16k/average_l0_123

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.22.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

Won

-0.50

Johan

-0.48

Pauline

-0.47

 Mauro

-0.46

 Jeon

-0.45

Gai

-0.44

 Pauline

-0.43

Mutant

-0.43

Jin

-0.43

Kl

-0.43

POSITIVE LOGITS

 address

1.13

 Address

1.12

address

1.02

ddress

1.00

 ADDRESS

1.00

Address

1.00

 getAddress

0.96

 addressed

0.93

 adress

0.92

 Addresses

0.91

Activations Density 0.071%

terms related to addresses in various formats

No Comments

No Known Activations