INDEX

Explanations

game-related information, possibly related to character skills and abilities

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

Topics

-0.80

LOG

-0.79

ylum

-0.78

é¾įå

-0.76

abilia

-0.75

itutional

-0.73

Gos

-0.71

forum

-0.71

 Brooks

-0.70

 Draper

-0.70

POSITIVE LOGITS

 projectiles

1.35

 inflic

1.33

 cooldown

1.32

AoE

1.32

 teleport

1.30

 heals

1.30

 inflict

1.29

 heal

1.23

 Piercing

1.22

 projectile

1.21

Activations Density 0.388%

game-related information, possibly related to character skills and abilities

No Comments

No Known Activations