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

 miracles

-0.76

 mosques

-0.70

 promoting

-0.69

 pharmacies

-0.69

 payday

-0.67

 legitim

-0.67

 promised

-0.66

 pharmacy

-0.65

 jihad

-0.64

 suites

-0.64

POSITIVE LOGITS

ocene

0.82

 Cellular

0.80

acus

0.77

ilial

0.75

ECK

0.73

imen

0.72

 Cricket

0.72

uca

0.71

cale

0.71

heter

0.71

Activations Density 0.000%

No Known Activations

This feature has no known activations.