INDEX

Explanations

words related to alleged events or accusations

oai_token-act-pair · gpt-3.5-turbo

references to allegations of wrongdoing or misconduct

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

oton

-0.82

uden

-0.80

APTER

-0.80

cair

-0.79

yards

-0.77

adan

-0.76

gaard

-0.75

fortunately

-0.74

ceans

-0.74

cloth

-0.73

POSITIVE LOGITS

 culprit

0.92

 perpetrator

0.88

 inability

0.87

 disappearance

0.87

 lack

0.85

 contradiction

0.85

 mastermind

0.85

 terrorist

0.84

 innocence

0.84

 threats

0.80

Activations Density 0.076%

words related to alleged events or accusations

references to allegations of wrongdoing or misconduct

No Comments

No Known Activations