INDEX

Explanations

phrases or terms referring to actions or situations that are considered unacceptable

oai_token-act-pair · gpt-3.5-turbo

expressions of unacceptability or dissatisfaction with certain behaviors or conditions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

tein

-0.90

ynthesis

-0.85

uay

-0.85

ulet

-0.84

ingers

-0.81

bard

-0.80

frey

-0.80

ools

-0.78

pel

-0.78

gio

-0.78

POSITIVE LOGITS

 unacceptable

0.90

 intolerable

0.82

 adolesc

0.78

 compromises

0.77

 undermin

0.73

 provocation

0.70

 degradation

0.69

 behaviour

0.68

 Ukrain

0.67

 srfAttach

0.67

Activations Density 0.014%

phrases or terms referring to actions or situations that are considered unacceptable

expressions of unacceptability or dissatisfaction with certain behaviors or conditions

No Comments

No Known Activations