INDEX

Explanations

specific phrases or titles from media and entertainment sources

oai_token-act-pair · gpt-3.5-turbo

titles or names of movies and their related elements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

anton

-0.84

tur

-0.80

ande

-0.78

private

-0.77

wal

-0.77

bub

-0.77

UC

-0.77

compan

-0.75

press

-0.74

manager

-0.72

POSITIVE LOGITS

 Thrones

1.03

 Pain

0.91

 Disaster

0.88

 Silence

0.86

 Violence

0.85

 Truth

0.82

 Influence

0.81

 Ashes

0.81

 Rails

0.81

 Feet

0.80

Activations Density 0.131%

specific phrases or titles from media and entertainment sources

titles or names of movies and their related elements

No Comments

No Known Activations