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

Negative Logits

cius

-0.70

erers

-0.68

hiba

-0.67

ãĥ£

-0.66

 Holden

-0.63

 Tuls

-0.60

Sold

-0.60

perty

-0.60

Hua

-0.60

Fug

-0.59

POSITIVE LOGITS

 anyway

1.09

 anyways

0.94

appropriate

0.73

oaded

0.69

asm

0.68

~~~~~~~~

0.68

MENTS

0.67

 Trojan

0.65

âķĲâķĲ

0.64

ascript

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.