INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

eu

-0.70

Ã§a

-0.69

à¨

-0.67

verb

-0.67

 witness

-0.65

utenant

-0.63

podcast

-0.63

wards

-0.63

ħĭ

-0.62

ening

-0.62

POSITIVE LOGITS

CFR

0.70

SpaceEngineers

0.69

imaru

0.65

ZI

0.65

she

0.64

é¾

0.64

pei

0.64

Zed

0.64

tone

0.63

ABE

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.