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

asin

-0.68

ators

-0.68

 point

-0.68

 neighbour

-0.66

ummer

-0.65

born

-0.64

 boil

-0.64

rical

-0.64

 toler

-0.63

 disadvantage

-0.62

POSITIVE LOGITS

Nib

0.78

Nit

0.74

 Thrones

0.71

 Enhancement

0.70

 Playoffs

0.69

 Featured

0.69

 Caps

0.68

 Contributions

0.67

 Squid

0.66

Sham

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.