INDEX

Explanations

mentions of Australian state abbreviations and government-related terms

oai_token-act-pair · gpt-3.5-turbo

references to geographic locations, specifically in relation to New South Wales (NSW) and the Australian Capital Territory (ACT)

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĥł

-0.80

bda

-0.80

toggle

-0.72

cean

-0.71

hower

-0.68

heimer

-0.68

akra

-0.67

aphael

-0.66

arters

-0.66

zai

-0.66

POSITIVE LOGITS

NSW

0.95

ADA

0.79

urst

0.78

UC

0.76

IRO

0.75

RL

0.75

CC

0.74

FW

0.74

DEF

0.71

 Defence

0.71

Activations Density 0.008%

mentions of Australian state abbreviations and government-related terms

references to geographic locations, specifically in relation to New South Wales (NSW) and the Australian Capital Territory (ACT)

No Comments

No Known Activations