INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 chunks

-0.65

 Bombs

-0.63

 Junk

-0.61

enegger

-0.59

 Decay

-0.59

 Malik

-0.58

 Thom

-0.58

 Bullet

-0.58

ront

-0.58

bill

-0.56

POSITIVE LOGITS

eret

0.78

ĪĴ

0.74

uable

0.73

particip

0.73

ysis

0.72

âĶģ

0.69

TextColor

0.67

âĵĺ

0.67

teness

0.67

orthy

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.