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

iversary

-0.78

govtrack

-0.74

 Ezek

-0.74

wine

-0.72

 Rica

-0.71

 raided

-0.70

aughters

-0.68

oun

-0.67

 Cummings

-0.67

Gur

-0.66

POSITIVE LOGITS

Shape

0.67

olt

0.64

 retweet

0.64

andowski

0.63

isphere

0.63

 Gutenberg

0.63

INST

0.62

formation

0.62

 interacting

0.62

 fielding

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.