INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

romeda

-0.79

-0.78

Mars

-0.73

Herm

-0.71

Excellent

-0.70

NOR

-0.69

bara

-0.68

hra

-0.66

WOOD

-0.66

Hill

-0.64

POSITIVE LOGITS

ank

0.80

inant

0.74

iliated

0.70

cript

0.69

cyclopedia

0.66

 coaster

0.64

avior

0.64

encing

0.63

 disag

0.62

osure

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.