INDEX

Explanations

instances of phrases related to observing or watching

oai_token-act-pair · gpt-3.5-turbo

phrases related to public visibility or accessibility

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

trap

-0.89

nesses

-0.67

redo

-0.66

pat

-0.66

working

-0.65

omp

-0.64

mph

-0.64

toe

-0.63

accompan

-0.63

alg

-0.63

POSITIVE LOGITS

 viewing

1.16

 "$:/

1.00

wat

0.86

ById

0.83

 experien

0.80

 screens

0.79

 Ratings

0.79

 views

0.76

 comprehension

0.74

 sessions

0.74

Activations Density 0.007%

instances of phrases related to observing or watching

phrases related to public visibility or accessibility

No Comments

No Known Activations