INDEX

Explanations

the word "scam"

oai_token-act-pair · gpt-4-turbo

references to various types of scams

oai_token-act-pair · gpt-3.5-turbo

references to scams

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

IFE

-0.68

itarian

-0.65

FK

-0.62

etheless

-0.61

LEASE

-0.60

arching

-0.60

 Views

-0.60

xc

-0.59

IGHTS

-0.59

 uninterrupted

-0.59

POSITIVE LOGITS

 scam

1.14

pering

1.03

ulent

1.03

ulence

1.02

crow

1.00

 scams

0.99

ming

0.92

sters

0.84

istry

0.84

roo

0.84

Activations Density 0.011%

the word "scam"

references to various types of scams

references to scams

No Comments

No Known Activations