INDEX

Explanations

phrases related to secret operatives or spies

oai_token-act-pair · gpt-3.5-turbo

terms related to covert agents or individuals involved in espionage or operations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ranged

-0.73

gets

-0.72

xual

-0.71

este

-0.71

yards

-0.70

bows

-0.70

gard

-0.68

Bah

-0.68

ŃĶ

-0.67

oran

-0.67

POSITIVE LOGITS

 operative

1.01

 operatives

0.80

eering

0.74

 Emails

0.72

sters

0.69

eers

0.68

 extraord

0.66

 Emanuel

0.66

èĢħ

0.65

doms

0.65

Activations Density 0.008%

phrases related to secret operatives or spies

terms related to covert agents or individuals involved in espionage or operations

No Comments

No Known Activations