INDEX

Explanations

No Explanations Found

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

 respons

-0.72

 Schwarz

-0.72

 Articles

-0.66

 reson

-0.65

 somet

-0.65

 favor

-0.65

 âĨĴ

-0.65

 probing

-0.64

Roe

-0.64

 favors

-0.62

POSITIVE LOGITS

get

1.89

ada

1.78

half

1.38

fits

1.04

gar

0.98

getic

0.98

getting

0.94

fit

0.94

mond

0.93

adas

0.92

Activations Density 0.000%

No Known Activations

This feature has no known activations.