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

orsi

-0.78

atches

-0.69

aments

-0.69

 sacrific

-0.66

lest

-0.65

 altar

-0.64

essee

-0.64

alty

-0.64

udeau

-0.63

oun

-0.63

POSITIVE LOGITS

PF

0.73

beh

0.72

ãĥ´

0.69

 promotion

0.67

Ms

0.62

MET

0.62

mouth

0.61

ba

0.60

fil

0.60

division

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.