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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

aunt

-0.66

enegger

-0.66

 Starg

-0.65

 Revelations

-0.65

 Witcher

-0.65

ãĥĩ

-0.64

terness

-0.64

{"

-0.63

 pornographic

-0.61

 browser

-0.60

POSITIVE LOGITS

iard

0.81

attach

0.74

onics

0.73

 spacing

0.69

 Leilan

0.68

proc

0.67

essel

0.65

bom

0.63

 stub

0.61

 nurs

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.