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

 Boots

-0.86

enhagen

-0.73

 skirts

-0.70

heit

-0.70

ONSORED

-0.69

oots

-0.68

ç«

-0.67

DATA

-0.64

 Gonzalez

-0.63

chini

-0.63

POSITIVE LOGITS

Mur

0.69

efully

0.67

ife

0.67

functional

0.67

 sober

0.65

esters

0.64

emonic

0.64

tesque

0.63

 deterrent

0.62

 haunted

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.