INDEX

Explanations

geographical locations, particularly directional references

oai_token-act-pair · gpt-3.5-turbo

directional or locational adjectives and modifiers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

otive

-0.81

Interstitial

-0.74

 VIDEOS

-0.73

leans

-0.72

Statement

-0.70

Fight

-0.68

rieved

-0.67

Column

-0.66

gage

-0.66

Ü

-0.66

POSITIVE LOGITS

 France

1.33

 Italy

1.29

 England

1.29

 Africa

1.28

 Germany

1.27

 Thailand

1.27

 Tanzania

1.27

 Spain

1.26

 Kenya

1.23

 Peru

1.22

Activations Density 0.083%

geographical locations, particularly directional references

directional or locational adjectives and modifiers

No Comments

No Known Activations