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

urst

-0.72

itizens

-0.70

Osc

-0.70

rack

-0.68

aneous

-0.67

akeru

-0.63

idered

-0.63

aders

-0.62

 Takeru

-0.61

 Scalia

-0.60

POSITIVE LOGITS

obal

0.68

atre

0.67

 impunity

0.67

eways

0.66

esian

0.62

arin

0.62

gewater

0.62

hedral

0.61

 Mahjong

0.61

builder

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.