INDEX

Explanations

mentions of violent acts resulting in death

oai_token-act-pair · gpt-3.5-turbo

references to death or fatalities

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

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

itory

-0.72

raft

-0.65

 Compass

-0.61

adjust

-0.61

TERN

-0.61

inker

-0.61

inking

-0.60

 trickle

-0.60

 Insight

-0.59

Uni

-0.59

POSITIVE LOGITS

 killed

3.46

 murdered

2.53

 slain

2.42

 Killed

2.42

killed

2.41

 slaughtered

2.38

 assassinated

2.33

 kills

2.23

 gunned

2.12

 kill

2.10

Activations Density 0.024%

mentions of violent acts resulting in death

references to death or fatalities

No Comments

No Known Activations