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

defense

-0.87

ynamic

-0.74

Defense

-0.74

ategory

-0.73

rius

-0.73

uzzle

-0.72

ynchronous

-0.70

cellaneous

-0.69

roller

-0.69

Downloadha

-0.67

POSITIVE LOGITS

 Isles

0.76

 understatement

0.75

oÄŁ

0.73

 Turks

0.71

 Antar

0.69

ONY

0.69

afort

0.68

 Gould

0.67

 Nieto

0.66

 Isle

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.