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

-1.08

)</

-0.79

encies

-0.76

earchers

-0.76

nai

-0.74

olin

-0.73

thren

-0.72

favorite

-0.70

Guide

-0.68

iasm

-0.68

POSITIVE LOGITS

 grabs

1.19

OTS

0.75

idy

0.71

plex

0.68

 Polo

0.66

 McMahon

0.63

 ObamaCare

0.62

 Reilly

0.62

cloth

0.61

HUD

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.