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

izont

-0.76

asio

-0.74

ennett

-0.73

ulhu

-0.69

orr

-0.69

rint

-0.68

zsche

-0.67

natureconservancy

-0.67

 bats

-0.67

users

-0.66

POSITIVE LOGITS

Sov

0.85

 Thai

0.71

BILITY

0.69

çĦ

0.67

LCS

0.66

 slideshow

0.66

èĢ

0.65

MENTS

0.63

onsense

0.61

 Scandinavian

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.