INDEX

Explanations

phrases suggesting an unwillingness to comment on a particular topic

oai_token-act-pair · gpt-3.5-turbo

instances of comments or statements regarding specific cases or events

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

fuck

-0.73

soDeliveryDate

-0.70

joice

-0.67

imilation

-0.65

ãĥī

-0.64

jam

-0.64

ynes

-0.63

 FranÃ§

-0.63

à¨

-0.63

KI

-0.63

POSITIVE LOGITS

 specifics

1.38

 whether

1.19

 particulars

1.14

 behalf

1.08

 specific

1.01

 pending

1.00

why

0.97

 details

0.95

 speculation

0.90

whether

0.90

Activations Density 0.095%

phrases suggesting an unwillingness to comment on a particular topic

instances of comments or statements regarding specific cases or events

No Comments

No Known Activations