INDEX

Explanations

mentions of violent actions and events

oai_token-act-pair · gpt-3.5-turbo

references to violence or gun-related incidents

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

erning

-0.82

 uniformly

-0.76

mittedly

-0.75

cedented

-0.70

etheless

-0.70

uitive

-0.70

 comprehens

-0.69

jri

-0.69

ctive

-0.69

 seamlessly

-0.68

POSITIVE LOGITS

 robbery

0.98

 robbers

0.97

 burgl

0.94

 raping

0.91

 robberies

0.90

killers

0.90

 gunshots

0.90

 stabbing

0.88

 killers

0.86

 killings

0.86

Activations Density 0.537%

mentions of violent actions and events

references to violence or gun-related incidents

No Comments

No Known Activations