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

cephal

-0.84

laus

-0.82

Ó

-0.78

esy

-0.70

aughs

-0.70

 Leban

-0.69

Emer

-0.69

orest

-0.69

 Bellev

-0.68

 Lanka

-0.68

POSITIVE LOGITS

LEASE

0.73

 targeted

0.67

MLA

0.66

 dumps

0.65

 procedural

0.63

imony

0.62

 dumping

0.62

 atomic

0.62

 apologies

0.62

 timet

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.