INDEX

Explanations

references to government or organizational offices

oai_token-act-pair · gpt-3.5-turbo

references to governmental or organizational offices

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

vous

-0.71

nir

-0.67

arget

-0.67

âķĲâķĲ

-0.65

itious

-0.64

uador

-0.63

icial

-0.63

unin

-0.62

MQ

-0.62

oise

-0.61

POSITIVE LOGITS

 Office

1.06

Office

0.94

 Depot

0.88

0.86

holders

0.81

office

0.80

holder

0.76

 Inspector

0.72

endar

0.71

velop

0.66

Activations Density 0.015%

references to government or organizational offices

references to governmental or organizational offices

No Comments

No Known Activations