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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

anche

-0.88

anches

-0.75

 organise

-0.71

 windows

-0.66

arios

-0.64

agu

-0.64

WAYS

-0.64

 whisk

-0.62

itaire

-0.62

 curtains

-0.61

POSITIVE LOGITS

©¶æ

0.83

NPR

0.81

 Collider

0.71

policy

0.68

NPR

0.66

lessness

0.62

specified

0.62

death

0.61

holder

0.61

 DACA

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.