INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

CPC

-0.80

(&

-0.72

Coy

-0.68

CTR

-0.66

 colleg

-0.62

anooga

-0.61

alion

-0.59

 Parish

-0.58

 Flame

-0.58

 Weir

-0.57

POSITIVE LOGITS

ugen

0.78

imet

0.76

itability

0.69

schild

0.69

imilar

0.65

paralleled

0.65

atown

0.65

orno

0.64

itable

0.63

uded

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.