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

addons

-0.75

 duct

-0.68

rug

-0.65

¶ħ

-0.62

 Pixie

-0.61

usha

-0.61

 conduc

-0.60

 maker

-0.58

ovie

-0.58

 Abandon

-0.58

POSITIVE LOGITS

Bib

0.74

amph

0.72

blems

0.72

eches

0.70

lehem

0.70

 colon

0.68

ithmetic

0.67

pperc

0.67

agues

0.67

ntil

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.