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

Ô

-0.79

Town

-0.75

nels

-0.72

;;;;;;;;;;;;

-0.72

 guiActiveUn

-0.69

neutral

-0.67

Charg

-0.65

kus

-0.65

');

-0.65

ĺħ

-0.64

POSITIVE LOGITS

VK

0.82

lip

0.68

 Adin

0.67

 Clover

0.64

 captcha

0.64

 Kash

0.64

 precedence

0.62

 Devi

0.61

ffee

0.59

 Sakuya

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.