INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 suppression

-0.71

bud

-0.69

pid

-0.68

pen

-0.68

sta

-0.66

 stru

-0.66

 ILCS

-0.66

jar

-0.65

chief

-0.64

hett

-0.64

POSITIVE LOGITS

 stripe

0.81

 Cosmos

0.73

 Hawking

0.72

 Generations

0.71

isode

0.70

 snipp

0.69

 Phant

0.68

 Helpful

0.67

 Britann

0.66

 Interstellar

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.