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

ãĥ¼ãĤ¯

-0.75

âĢ¢âĢ¢

-0.70

Raw

-0.69

 Flames

-0.69

kus

-0.69

DVD

-0.66

punk

-0.66

bars

-0.66

ragon

-0.65

 synd

-0.64

POSITIVE LOGITS

gew

0.72

 conduc

0.71

upp

0.68

 admitting

0.65

onym

0.64

imer

0.64

 recess

0.63

aspers

0.63

 sidel

0.63

cknowled

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.