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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

accompan

-0.92

axter

-0.78

 suscept

-0.75

arily

-0.75

////////////////////////////////

-0.74

byss

-0.70

 assassinated

-0.69

leck

-0.68

ModLoader

-0.68

 snipp

-0.68

POSITIVE LOGITS

 Google

1.46

 google

1.17

Google

1.05

phabet

0.85

 search

0.76

 fire

0.74

 Youtube

0.72

ument

0.71

google

0.70

 curiosity

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.