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

Mub

-0.67

 Accessed

-0.66

il

-0.61

 responsive

-0.60

 Mole

-0.60

 wall

-0.60

Zy

-0.59

 marsh

-0.59

 tolerance

-0.59

 Nile

-0.58

POSITIVE LOGITS

DCS

0.88

Ö¼

0.83

redo

0.78

\\\\\\\\

0.78

paio

0.75

helm

0.74

ADRA

0.73

inelli

0.73

emort

0.71

-+-+

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.