INDEX

Explanations

terms related to geographical locations, particularly North and South Korea

oai_token-act-pair · gpt-3.5-turbo

mentions of North and South Korea

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

iddler

-0.72

 rich

-0.67

 Thread

-0.67

 Ryder

-0.63

 feat

-0.62

Kot

-0.62

tab

-0.61

 richer

-0.60

 thriving

-0.60

Coh

-0.60

POSITIVE LOGITS

west

1.13

withstanding

0.98

wards

0.96

etheless

0.96

theless

0.95

agascar

0.95

mosp

0.94

alyst

0.93

bidden

0.92

xiety

0.92

Activations Density 0.012%

terms related to geographical locations, particularly North and South Korea

mentions of North and South Korea

No Comments

No Known Activations