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

 Tinker

-0.67

 Robotics

-0.67

 Cost

-0.65

uits

-0.65

catentry

-0.64

 Fees

-0.63

osaurs

-0.63

è¦ļéĨĴ

-0.62

 Genius

-0.61

 Norn

-0.61

POSITIVE LOGITS

 overboard

0.71

 fortunately

0.70

 hurd

0.68

alsh

0.67

 plent

0.64

 desper

0.63

 ende

0.63

xual

0.63

roth

0.63

MG

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.