INDEX

Explanations

words related to confidential or secret information

oai_token-act-pair · gpt-3.5-turbo

references to classified information and related concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

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

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

 redeemed

-0.75

Kay

-0.73

arcity

-0.70

abeth

-0.69

annis

-0.68

llan

-0.67

galitarian

-0.67

raught

-0.67

ador

-0.67

alach

-0.66

POSITIVE LOGITS

 information

1.45

 briefings

1.37

 documents

1.30

 material

1.22

 info

1.19

 materials

1.14

 memos

1.10

 intelligence

1.09

 INFORMATION

1.09

 intel

1.08

Activations Density 0.070%

words related to confidential or secret information

references to classified information and related concepts

No Comments

No Known Activations