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

eng

-0.83

abel

-0.74

pas

-0.74

 Presidency

-0.72

ote

-0.70

ector

-0.69

ree

-0.69

ocalypse

-0.69

etimes

-0.69

ypes

-0.68

POSITIVE LOGITS

maxwell

0.74

 cloudy

0.67

 sour

0.66

 cowork

0.66

 reeling

0.66

PASS

0.65

 humming

0.65

CRA

0.65

 veterinarian

0.64

ecd

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.