INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ilater

-0.75

oute

-0.70

arij

-0.70

anut

-0.67

foundland

-0.67

ibliography

-0.65

ruary

-0.64

igraph

-0.64

scl

-0.63

io

-0.62

POSITIVE LOGITS

è¦

0.64

anted

0.63

voy

0.62

 accord

0.61

 dolphins

0.60

 valued

0.59

 pony

0.59

 feat

0.59

 matched

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.