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

 mapping

-0.67

 math

-0.65

 diving

-0.62

span

-0.60

 maths

-0.59

rom

-0.58

 reading

-0.58

feeding

-0.57

leading

-0.57

bar

-0.57

POSITIVE LOGITS

antine

0.78

oled

0.72

oka

0.71

phies

0.69

aughters

0.69

NetMessage

0.68

OTOS

0.65

ikk

0.64

gemony

0.64

ogle

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.