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

etheless

-0.87

oodle

-0.68

 plur

-0.68

umbn

-0.68

 constitu

-0.66

 ther

-0.65

 cybersecurity

-0.64

Reviewer

-0.64

 princ

-0.63

ename

-0.62

POSITIVE LOGITS

soDeliveryDate

0.76

ulus

0.73

overty

0.67

ersion

0.65

EMENT

0.63

 Efficiency

0.61

ULAR

0.60

am

0.60

inance

0.59

abal

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.