INDEX

Explanations

negative actions and situations involving people

oai_token-act-pair · gpt-3.5-turbo

expressions of emotional states and responses to conflict

oai_token-act-pair · gpt-4o-mini Triggered by @bot

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

ãĤ´ãĥ³

-1.07

BuyableInstoreAndOnline

-0.92

 nostalg

-0.82

olkien

-0.80

 Flavoring

-0.78

 Tolkien

-0.77

atari

-0.74

 Achievements

-0.73

optim

-0.73

iscover

-0.73

POSITIVE LOGITS

 tresp

1.24

 robbery

1.24

 cops

1.22

 burglary

1.21

police

1.18

 gunshots

1.15

 handcuffs

1.14

 SWAT

1.14

 handcuffed

1.14

 burgl

1.13

Activations Density 0.659%

negative actions and situations involving people

expressions of emotional states and responses to conflict

No Comments

No Known Activations