INDEX

Explanations

references to security and intelligence agencies and their activities

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

phrases related to surveillance and intelligence agencies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_post

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Sadd

-0.73

aez

-0.73

stick

-0.71

 redeemed

-0.71

ãĤ©

-0.69

rene

-0.67

 Wooden

-0.67

sticks

-0.66

Pain

-0.66

Ry

-0.65

POSITIVE LOGITS

spy

1.53

 spying

1.52

 spies

1.49

 eaves

1.46

 intelligence

1.43

 espionage

1.40

 interception

1.37

CHQ

1.36

NSA

1.34

 surveillance

1.32

Activations Density 0.398%

references to security and intelligence agencies and their activities

phrases related to surveillance and intelligence agencies

No Comments

No Known Activations