INDEX

Explanations

instances of the word "are" preceded by various subjects

oai_token-act-pair · gpt-3.5-turbo

the end of a document or a significant cessation of content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

icism

-0.75

 Azerb

-0.72

 agre

-0.66

 Seym

-0.66

icter

-0.65

smanship

-0.63

 Niet

-0.63

Redditor

-0.59

dom

-0.59

 lapt

-0.58

POSITIVE LOGITS

wolves

0.95

wolf

0.87

senal

0.80

 SHARES

0.73

types

0.66

 photos

0.63

 hereby

0.61

ctions

0.60

Ĥª

0.60

 guest

0.56

Activations Density 0.058%

instances of the word "are" preceded by various subjects

the end of a document or a significant cessation of content

No Comments

No Known Activations