INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

AFTA

-0.90

 plaintiffs

-0.70

 actors

-0.64

?????-

-0.63

 defendants

-0.63

 considerations

-0.63

Preview

-0.63

 prospects

-0.61

 provisions

-0.61

ocket

-0.60

POSITIVE LOGITS

 contrace

0.70

 Maiden

0.67

 Swanson

0.67

 Grail

0.67

 Twins

0.67

 Monarch

0.67

 Frie

0.66

 whence

0.64

 Glover

0.64

 Purg

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.