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

wordpress

-0.81

 collapses

-0.68

 asbestos

-0.68

azeera

-0.68

gnu

-0.67

mberg

-0.66

imgur

-0.66

ucl

-0.65

hin

-0.65

asca

-0.64

POSITIVE LOGITS

 preserve

0.71

 gain

0.71

Queue

0.71

Redditor

0.69

 keep

0.68

Vote

0.68

racuse

0.67

OOL

0.65

Override

0.65

 adopt

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.