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

 Bulg

-0.71

 Leilan

-0.66

 Blackhawks

-0.65

needed

-0.64

 refres

-0.64

uably

-0.63

iod

-0.63

ories

-0.63

orsi

-0.62

POSITIVE LOGITS

quart

0.68

 otherwise

0.68

ô

0.67

und

0.64

GoldMagikarp

0.64

pain

0.63

 under

0.60

 abusive

0.60

 tread

0.59

 undert

0.57

Activations Density 0.000%

No Known Activations

This feature has no known activations.