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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 millenn

-0.84

ICA

-0.78

SEC

-0.73

Ô

-0.71

 Palestin

-0.71

 unbeliev

-0.69

 comr

-0.69

Measure

-0.68

Civil

-0.68

Spl

-0.67

POSITIVE LOGITS

ð

0.80

lihood

0.80

EngineDebug

0.78

wcsstore

0.76

veyard

0.73

dropping

0.73

ikan

0.72

rentice

0.72

mare

0.70

"},"

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.