INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 rupt

-0.76

galitarian

-0.75

artifacts

-0.74

 democrat

-0.71

 Scrolls

-0.62

 precip

-0.62

NetMessage

-0.61

 doomed

-0.60

wcsstore

-0.59

 liqu

-0.59

POSITIVE LOGITS

 Collider

0.83

 Berm

0.73

iyah

0.68

 Rapp

0.67

 Cage

0.67

ounced

0.66

rador

0.62

anamo

0.62

 Review

0.61

lance

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.