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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

urses

-0.69

 charism

-0.67

 Osaka

-0.65

 humid

-0.63

jun

-0.63

 Lauder

-0.62

 fortunes

-0.62

 stash

-0.61

ori

-0.61

 stray

-0.59

POSITIVE LOGITS

adium

0.83

tones

0.78

bucks

0.70

Artist

0.70

die

0.69

bard

0.67

Virgin

0.66

 Werewolf

0.65

0.62

peak

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.