INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ooming

-0.74

vir

-0.68

rontal

-0.67

unal

-0.66

irez

-0.66

odium

-0.65

iod

-0.65

EV

-0.64

ulhu

-0.63

agra

-0.63

POSITIVE LOGITS

 Schumer

0.73

 newsletters

0.69

(@

0.64

ault

0.64

 Barkley

0.63

 Beck

0.62

 Kass

0.62

 Jenn

0.61

essional

0.61

 Cambodia

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.