INDEX

Explanations

names or words containing the letters "er" followed by a digit

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals or their roles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

-0.81

scl

-0.76

ĪĴ

-0.72

 Lovecraft

-0.66

ersen

-0.63

-0.62

âĢ¢âĢ¢

-0.60

ertodd

-0.60

 parap

-0.59

ELS

-0.57

POSITIVE LOGITS

jee

1.17

adish

1.03

nery

0.98

idge

0.96

unning

0.88

usalem

0.88

lein

0.86

mens

0.85

ich

0.83

ickson

0.83

Activations Density 0.064%

names or words containing the letters "er" followed by a digit

references to specific individuals or their roles

No Comments

No Known Activations