INDEX

Explanations

texts related to spam and potential scams

oai_token-act-pair · gpt-3.5-turbo

references to spam and scams in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

hani

-1.10

 Borders

-0.73

 cider

-0.72

ederal

-0.69

IFE

-0.66

 Dayton

-0.64

 Judges

-0.63

 Differences

-0.63

 Difference

-0.63

 Governors

-0.62

POSITIVE LOGITS

ming

1.42

mer

1.23

mers

1.02

ulent

1.00

pering

1.00

crow

0.99

my

0.95

pling

0.91

vertising

0.89

bara

0.89

Activations Density 0.074%

texts related to spam and potential scams

references to spam and scams in the text

No Comments

No Known Activations