INDEX

Explanations

references to legal or disciplinary actions

oai_token-act-pair · gpt-3.5-turbo

mentions of punishment in various contexts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

eds

-0.82

NetMessage

-0.76

soDeliveryDate

-0.75

sonian

-0.72

rote

-0.72

ergy

-0.71

roots

-0.69

zyme

-0.69

gow

-0.67

thus

-0.67

POSITIVE LOGITS

 punishment

1.18

 punishments

1.05

 punished

1.03

 punish

1.02

 lashes

0.90

 inflicted

0.87

 sanction

0.85

 harshly

0.84

 penalty

0.84

 torment

0.83

Activations Density 0.015%

references to legal or disciplinary actions

mentions of punishment in various contexts

No Comments

No Known Activations