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

hower

-0.84

TPS

-0.74

 Atkins

-0.68

 Seym

-0.65

æ©Ł

-0.65

iband

-0.65

CHO

-0.64

 radios

-0.64

 releg

-0.64

yip

-0.62

POSITIVE LOGITS

enes

0.81

pell

0.80

anmar

0.75

irst

0.72

nil

0.71

description

0.71

abol

0.70

este

0.68

athed

0.67

rings

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.