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

Reilly

-0.75

yip

-0.71

 pageant

-0.69

Sab

-0.67

TT

-0.67

Miss

-0.67

Jam

-0.67

 Chattanooga

-0.66

ANE

-0.66

 Tess

-0.65

POSITIVE LOGITS

 prescribed

0.77

unci

0.72

 prud

0.72

.}

0.70

lied

0.70

 doctrine

0.70

 heirs

0.70

atorium

0.69

cius

0.69

acterial

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.