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

risome

-0.73

 substitute

-0.66

Ã¨re

-0.65

actionDate

-0.62

 acquaint

-0.61

 Respect

-0.60

ECO

-0.60

 Reconstruction

-0.60

 Consider

-0.59

 Temper

-0.59

POSITIVE LOGITS

0.76

Eva

0.72

bank

0.72

cedes

0.70

Accessory

0.70

veyard

0.67

tery

0.67

Tex

0.65

ahu

0.65

exec

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.