INDEX

Explanations

mentions of specific geographic locations like states, countries, and homelands

oai_token-act-pair · gpt-3.5-turbo

terms related to geographical locations and political borders

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

inventoryQuantity

-0.83

Management

-0.73

Prev

-0.73

Chair

-0.69

Hour

-0.68

thumbnails

-0.68

Writ

-0.67

 Emails

-0.67

soDeliveryDate

-0.66

 McConnell

-0.65

POSITIVE LOGITS

 nationals

1.07

 borders

1.00

istan

0.99

 annexed

0.97

 countries

0.96

 cuisine

0.95

 nations

0.95

 ravaged

0.90

 shores

0.87

 economies

0.86

Activations Density 0.374%

mentions of specific geographic locations like states, countries, and homelands

terms related to geographical locations and political borders

No Comments

No Known Activations