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

 Prescott

-0.85

 Goff

-0.78

chip

-0.72

erity

-0.70

value

-0.65

 Darwin

-0.64

Factor

-0.64

Dak

-0.63

fun

-0.63

kill

-0.63

POSITIVE LOGITS

theless

0.85

agan

0.78

otos

0.77

withstanding

0.74

icans

0.73

united

0.68

etheless

0.68

 swayed

0.67

enza

0.67

 contradicted

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.