INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ĸļ

-0.83

 inexper

-0.76

utenberg

-0.73

 instincts

-0.71

 Painter

-0.65

icz

-0.63

Fuj

-0.62

 surgery

-0.62

 spores

-0.62

idth

-0.62

POSITIVE LOGITS

mun

0.84

BILL

0.78

apo

0.77

Portland

0.76

nown

0.73

shall

0.72

TAG

0.72

ociated

0.72

ingham

0.71

Americ

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.