INDEX

Explanations

phrases related to professional practices or activities

oai_token-act-pair · gpt-3.5-turbo

instances of time, duration, and references to practice or performance situations

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

kefeller

-0.78

arnaev

-0.71

Detailed

-0.68

osponsors

-0.67

¥ŀ

-0.64

groupon

-0.60

ongyang

-0.60

ãĤ´ãĥ³

-0.58

SourceFile

-0.58

ogly

-0.57

POSITIVE LOGITS

1.30

we

1.19

 they

1.03

 nobody

1.00

you

0.99

 yeah

0.96

 though

0.95

 everybody

0.93

it

0.93

he

0.93

Activations Density 0.255%

phrases related to professional practices or activities

instances of time, duration, and references to practice or performance situations

No Comments

No Known Activations