INDEX

Explanations

phrases indicating actions related to printing or physically manipulating something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "it" and variations of its usage

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

idth

-0.73

 Frontier

-0.68

 Polk

-0.60

ãĥ©ãĥ³

-0.59

avage

-0.56

ãĤ¢

-0.55

"],"

-0.55

 Vehicle

-0.55

 Confeder

-0.54

²¾

-0.54

POSITIVE LOGITS

alian

1.41

self

1.23

unes

1.21

chy

0.96

iner

0.89

zbollah

0.85

selves

0.82

 aloud

0.80

 horizont

0.80

atic

0.78

Activations Density 0.155%

phrases indicating actions related to printing or physically manipulating something

instances of the word "it" and variations of its usage

No Comments

No Known Activations