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

irez

-0.81

aga

-0.70

@#&

-0.70

################################

-0.69

 Vert

-0.69

rael

-0.68

liam

-0.67

nder

-0.66

awaru

-0.64

eu

-0.64

POSITIVE LOGITS

 Lauder

0.68

 disclosing

0.65

limits

0.64

fin

0.63

 aperture

0.61

women

0.60

 sensitivity

0.60

fighters

0.60

 calculations

0.60

 aids

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.