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

uggle

-0.71

 blame

-0.69

overe

-0.67

umbnail

-0.64

orthy

-0.63

atial

-0.63

lease

-0.63

ulner

-0.62

 swat

-0.61

angered

-0.61

POSITIVE LOGITS

utor

0.73

 Riders

0.72

 Chocobo

0.72

bike

0.70

 rider

0.69

 Pony

0.68

CLUD

0.67

Cart

0.66

 pony

0.66

Ton

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.