INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

theless

-0.86

 DEFENSE

-0.74

abulary

-0.68

gur

-0.68

istar

-0.68

 wrestling

-0.68

chin

-0.68

atta

-0.67

angan

-0.65

apon

-0.64

POSITIVE LOGITS

wagen

0.74

ãĥ´

0.69

+)

0.63

 emulation

0.63

ãĤ»

0.63

Recomm

0.63

âĻ

0.62

ãĥĩãĤ£

0.61

usal

0.60

install

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.