INDEX

Explanations

references to sacred or sacrilegious concepts

oai_token-act-pair · gpt-3.5-turbo

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ously

-0.77

ITNESS

-0.76

çīĪ

-0.75

DonaldTrump

-0.72

 Clarkson

-0.69

 Sawyer

-0.67

 Palmer

-0.65

EEP

-0.63

 Philips

-0.63

!/

-0.62

POSITIVE LOGITS

rament

1.08

het

1.07

ificial

1.03

cer

0.97

cery

0.95

rum

0.94

culus

0.94

hem

0.90

char

0.90

hest

0.89

Activations Density 0.083%

references to sacred or sacrilegious concepts

No Comments

No Known Activations