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

chery

-0.74

taboola

-0.69

allery

-0.69

 Gulf

-0.67

Yel

-0.66

glers

-0.66

thora

-0.64

irlf

-0.63

åĨ

-0.63

Gul

-0.63

POSITIVE LOGITS

Emb

0.80

mie

0.77

Ne

0.68

ine

0.64

aps

0.63

 Lange

0.63

iqueness

0.63

ise

0.62

EA

0.62

iken

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.