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

 Canaver

-0.77

emale

-0.73

estinal

-0.73

uclear

-0.72

orc

-0.72

armac

-0.71

iple

-0.69

uminati

-0.69

ixt

-0.69

ulton

-0.68

POSITIVE LOGITS

FTWARE

0.74

 Marriott

0.63

 exception

0.63

izing

0.59

 nature

0.59

isSpecialOrderable

0.59

 crow

0.58

DOM

0.58

 intellig

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.