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

munition

-0.73

aundering

-0.70

 match

-0.64

 Wrest

-0.62

 toget

-0.61

ilet

-0.60

avorite

-0.59

 grease

-0.59

 ende

-0.59

 transplant

-0.57

POSITIVE LOGITS

amaz

0.80

Frames

0.78

gs

0.72

opens

0.71

pic

0.68

onica

0.67

join

0.67

fml

0.67

ris

0.65

Frameworks

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.