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

PLA

-0.77

REG

-0.77

NRS

-0.70

âĢ¦)

-0.70

×Ļ×

-0.68

ALE

-0.66

DES

-0.66

————————

-0.64

PRO

-0.63

asonic

-0.63

POSITIVE LOGITS

ogle

0.79

olor

0.75

emark

0.72

iton

0.72

etheless

0.72

onet

0.71

oplan

0.70

assian

0.69

leck

0.69

rued

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.