INDEX

Explanations

word forms related to committing or intending to commit actions that are negative or have legal consequences

oai_token-act-pair · gpt-3.5-turbo

phrases associated with the act of committing various offenses or illegal actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

 complexion

-0.72

adish

-0.68

dayName

-0.66

sama

-0.66

atin

-0.66

 Flavoring

-0.65

WARN

-0.62

mare

-0.61

bowl

-0.60

framework

-0.60

POSITIVE LOGITS

 suicide

1.00

 adultery

0.85

 committed

0.83

 committing

0.82

utions

0.80

acts

0.79

 commit

0.75

icts

0.74

ruction

0.73

 offences

0.73

Activations Density 0.025%

word forms related to committing or intending to commit actions that are negative or have legal consequences

phrases associated with the act of committing various offenses or illegal actions

No Comments

No Known Activations