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

ought

-0.76

luck

-0.71

Hoo

-0.68

velt

-0.67

Peb

-0.65

 Duchess

-0.63

uben

-0.62

Tav

-0.62

 Paradox

-0.61

 Golem

-0.61

POSITIVE LOGITS

 enthusi

0.72

enum

0.68

mortem

0.67

 disse

0.66

ãĥīãĥ©

0.63

RY

0.63

 governance

0.62

 microscope

0.62

ãĥĩ

0.61

senal

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.