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

Ri

-0.70

assi

-0.67

Fat

-0.64

 toast

-0.64

 Ethiopian

-0.63

Peb

-0.63

awa

-0.63

 prox

-0.62

auna

-0.61

 Turks

-0.61

POSITIVE LOGITS

paying

0.88

dated

0.78

paid

0.74

license

0.70

 Contracts

0.68

 Casting

0.66

Dead

0.64

letters

0.62

hazard

0.62

union

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.