INDEX

Explanations

phrases related to confidentiality and secrets

oai_token-act-pair · gpt-3.5-turbo

instances of the word "confidential" and related terms

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

annis

-0.95

Ì¶

-0.84

akeru

-0.75

alis

-0.75

plex

-0.74

hao

-0.73

slow

-0.73

iph

-0.73

ixels

-0.72

uph

-0.71

POSITIVE LOGITS

 confidential

1.28

 informant

1.13

idential

1.03

 informants

1.02

 correspondence

0.87

 confidentiality

0.85

 memos

0.82

 documents

0.81

ensable

0.80

 information

0.78

Activations Density 0.010%

phrases related to confidentiality and secrets

instances of the word "confidential" and related terms

No Comments

No Known Activations