INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 robotics

-0.75

 charact

-0.71

 connectivity

-0.63

 portfolio

-0.62

[]

-0.62

 favourites

-0.62

 licens

-0.61

 arrays

-0.61

 geometry

-0.61

 favourite

-0.61

POSITIVE LOGITS

Bir

0.79

edom

0.79

oros

0.72

ĪĴ

0.70

Church

0.70

perty

0.69

nih

0.69

heid

0.69

transfer

0.68

ticket

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.