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

Category

-0.78

-0.77

66666666

-0.75

Redditor

-0.73

Artist

-0.70

 guiName

-0.67

racuse

-0.67

Iter

-0.66

 BASE

-0.64

 quantities

-0.64

POSITIVE LOGITS

claw

0.81

aundering

0.76

otics

0.74

eties

0.73

roth

0.71

oth

0.69

iatrics

0.68

love

0.67

ionics

0.67

nos

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.