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

nesday

-0.74

éĹĺ

-0.72

lished

-0.72

ERY

-0.70

uers

-0.69

mble

-0.68

CHR

-0.67

eways

-0.66

WHERE

-0.65

lly

-0.65

POSITIVE LOGITS

 bombard

0.73

 Battalion

0.68

 Survive

0.67

 redd

0.66

 Underworld

0.66

 Colony

0.66

 Commando

0.64

 Pill

0.63

redd

0.63

 Squadron

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.