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

%]

-0.71

Rating

-0.69

 Ribbon

-0.68

alist

-0.67

ħĭ

-0.65

KB

-0.62

ebted

-0.62

eret

-0.62

bernatorial

-0.60

OTAL

-0.60

POSITIVE LOGITS

agents

0.69

iants

0.67

estic

0.67

 Zoro

0.65

 foss

0.64

zo

0.63

ython

0.63

raf

0.62

Jav

0.62

asma

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.