INDEX

Explanations

action verbs, particularly variations of "do."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 houſe

-0.89

Ise

-0.83

Portail

-0.82

 ſtate

-0.78

 ſche

-0.75

 perſon

-0.75

 rodríguez

-0.74

 scattata

-0.74

 fernández

-0.74

 sánchez

-0.74

POSITIVE LOGITS

 done

1.72

Doing

1.46

 doing

1.44

do

1.40

doing

1.40

 Doing

1.36

 DOING

1.34

done

1.30

 DONE

1.29

 doin

1.28

Activations Density 0.110%

action verbs, particularly variations of "do."

No Comments

No Known Activations