INDEX

Explanations

phrases related to politics and power dynamics

oai_token-act-pair · gpt-3.5-turbo

phrases related to intelligence and critical thinking

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

orer

-0.69

TION

-0.68

idae

-0.67

escription

-0.66

ccording

-0.65

Driver

-0.65

payment

-0.64

¤

-0.64

Leod

-0.62

 Previous

-0.62

POSITIVE LOGITS

 halls

1.88

 offices

1.71

 corridors

1.70

 classrooms

1.63

 rooms

1.60

 desks

1.53

 capitals

1.53

 bedrooms

1.50

 kitchens

1.48

 chambers

1.45

Activations Density 0.572%

phrases related to politics and power dynamics

phrases related to intelligence and critical thinking

No Comments

No Known Activations