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

ports

-0.64

ual

-0.63

 torn

-0.60

 advent

-0.58

 proud

-0.58

çĲ

-0.58

 establishment

-0.57

UT

-0.57

urance

-0.57

 stretching

-0.56

POSITIVE LOGITS

ħĭ

0.77

iddles

0.71

aily

0.71

schild

0.71

 citiz

0.71

DonaldTrump

0.71

olding

0.69

 Peng

0.69

ibling

0.69

 Gentleman

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.