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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ipes

-0.74

 Madden

-0.67

 Dise

-0.67

Raven

-0.66

 Strauss

-0.65

 Puzzles

-0.62

 oranges

-0.61

asty

-0.60

swe

-0.60

iles

-0.59

POSITIVE LOGITS

displayText

0.88

SPONSORED

0.77

cedented

0.70

aud

0.69

ãĤ¸

0.67

mentioned

0.66

redited

0.65

tesy

0.65

 Sharif

0.64

Marginal

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.