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

 Mechdragon

-0.85

hov

-0.77

 Saiyan

-0.76

Ranked

-0.70

 Shroud

-0.70

Prince

-0.67

steen

-0.67

rawdownloadcloneembedreportprint

-0.66

 Ahead

-0.66

pping

-0.64

POSITIVE LOGITS

osite

0.82

swer

0.74

isphere

0.72

weather

0.71

:/

0.71

igators

0.67

ittee

0.66

onite

0.66

cn

0.66

atcher

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.