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

manship

-0.75

 Gott

-0.73

WER

-0.70

erb

-0.66

Sir

-0.65

SO

-0.65

MX

-0.64

olas

-0.63

Ud

-0.63

aughs

-0.63

POSITIVE LOGITS

 exclusive

1.52

exclusive

1.25

 Exclusive

1.06

avored

0.78

ãĥ¼ãĥ³

0.77

angan

0.74

Ń·

0.71

ibrary

0.71

ensing

0.71

taboola

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.