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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

icum

-0.72

 Printed

-0.69

ribing

-0.68

lamm

-0.68

 recip

-0.64

culation

-0.63

mitted

-0.62

âĸĪâĸĪâĸĪâĸĪâĸĪâĸĪâĸĪâĸĪ

-0.61

rency

-0.60

 mosqu

-0.60

POSITIVE LOGITS

Na

0.75

Cu

0.69

 Kane

0.68

Hue

0.67

fare

0.67

 Kung

0.66

Kak

0.66

 Liter

0.62

 Kick

0.62

 Robo

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.