INDEX

Explanations

the word "la" with a strong activation value

oai_token-act-pair · gpt-3.5-turbo

repetitive sound patterns or phrases

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 remed

-0.68

SPD

-0.65

 Investigative

-0.64

ãĥ¡

-0.63

 scrub

-0.63

 Samurai

-0.63

 discern

-0.62

 rodent

-0.58

 neurot

-0.58

 browsing

-0.58

POSITIVE LOGITS

la

4.39

lah

1.95

las

1.77

lo

1.55

li

1.54

lan

1.47

LA

1.44

La

1.38

lia

1.35

lar

1.32

Activations Density 0.013%

the word "la" with a strong activation value

repetitive sound patterns or phrases

No Comments

No Known Activations