INDEX

Explanations

male and female gender-specific terms

oai_token-act-pair · gpt-3.5-turbo

references to male and female identities or roles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

èª

-0.82

ENCY

-0.71

ģĸ

-0.71

ENTS

-0.69

OPS

-0.68

ears

-0.67

{{

-0.65

NRS

-0.65

ãĥĦ

-0.64

î

-0.64

POSITIVE LOGITS

volent

1.25

 Male

1.15

 Female

0.94

 Males

0.79

vich

0.79

uscript

0.73

ering

0.73

 genital

0.73

 mating

0.68

Ath

0.65

Activations Density 0.007%

male and female gender-specific terms

references to male and female identities or roles

No Comments

No Known Activations