INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

oppable

-0.77

geries

-0.77

ACP

-0.73

elaide

-0.73

ļéĨĴ

-0.72

regor

-0.71

 constitu

-0.69

vertisements

-0.68

GV

-0.67

ificant

-0.66

POSITIVE LOGITS

 wings

0.78

 Hazard

0.68

 Rossi

0.66

Hai

0.66

":[{"

0.65

 Dynam

0.64

wings

0.64

 Duff

0.64

":[

0.64

Raz

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.