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

phabet

-0.76

det

-0.76

Obj

-0.72

uyomi

-0.69

iga

-0.68

 collide

-0.65

ql

-0.65

 Rasmussen

-0.65

bourg

-0.64

odan

-0.62

POSITIVE LOGITS

Ms

0.76

 Sett

0.73

HP

0.67

 Slave

0.65

idges

0.64

 Sect

0.64

School

0.63

YING

0.61

HER

0.61

ESE

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.