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

til

-0.84

angle

-0.83

angles

-0.78

Dia

-0.76

Cooldown

-0.70

Interstitial

-0.69

split

-0.67

minute

-0.67

peak

-0.66

 tilt

-0.66

POSITIVE LOGITS

 Mehran

0.86

itatively

0.76

 Butcher

0.68

 cultiv

0.64

 stockp

0.63

ilib

0.63

ollen

0.62

oples

0.62

 beaut

0.61

 Rhino

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.