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

ullivan

-0.76

 Donation

-0.69

ived

-0.67

 Â·

-0.67

 Skies

-0.66

rats

-0.65

 Â»

-0.63

clud

-0.63

-->

-0.61

iked

-0.61

POSITIVE LOGITS

âĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢ

0.70

 apologise

0.68

Charge

0.68

Wem

0.67

OTAL

0.66

 simultane

0.65

 pronouns

0.64

 pacif

0.63

 Collider

0.63

mins

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.