INDEX

Explanations

mentions of specific names and terms like "MacArthur," "Bernardino," and "DACA."

oai_token-act-pair · gpt-3.5-turbo

proper nouns and terms related to specific locations, events, or organizations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

hend

-0.76

 combustion

-0.76

dust

-0.71

Steel

-0.71

Truth

-0.69

 Quantum

-0.68

 futuristic

-0.67

 Fired

-0.66

nit

-0.66

Engine

-0.65

POSITIVE LOGITS

 Bernardino

2.22

 MacArthur

1.83

 DACA

1.24

 Byrne

1.21

 Disneyland

1.15

 kindred

1.00

gap

0.99

 Gomez

0.98

 rainfall

0.96

fol

0.96

Activations Density 0.065%

mentions of specific names and terms like "MacArthur," "Bernardino," and "DACA."

proper nouns and terms related to specific locations, events, or organizations

No Comments

No Known Activations