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

 diagonal

-0.76

Eag

-0.75

 subdivision

-0.67

hered

-0.65

 braking

-0.61

 Sapp

-0.61

 Mong

-0.60

 exile

-0.60

 savings

-0.59

 Soros

-0.59

POSITIVE LOGITS

ologist

0.82

UTH

0.78

track

0.77

illus

0.76

lead

0.75

lore

0.73

san

0.73

aris

0.73

Lead

0.71

ology

0.71

Activations Density 0.000%

No Known Activations

This feature has no known activations.