INDEX

Explanations

the word "scam" and its variations, such as "scamming" and "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 @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

-0.73

FK

-0.71

IFE

-0.67

itarian

-0.63

 Atom

-0.63

 Differences

-0.62

UTC

-0.61

 Olive

-0.60

éĹĺ

-0.57

 Idle

-0.56

POSITIVE LOGITS

ulent

1.23

ulence

1.07

pering

1.03

ming

0.98

sters

0.96

ulously

0.91

crow

0.91

 scams

0.88

 scam

0.88

atown

0.85

Activations Density 0.018%

the word "scam" and its variations, such as "scamming" and "scams."

references to scams

No Comments

No Known Activations