INDEX

Explanations

mentions of the word "pie"

oai_token-act-pair · gpt-3.5-turbo

references to pie

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

eness

-0.75

LR

-0.68

 Registrar

-0.67

HB

-0.65

ih

-0.65

 Advis

-0.64

 Avoid

-0.63

 Cour

-0.63

 Affect

-0.62

Guard

-0.62

POSITIVE LOGITS

pie

4.19

Pie

2.81

 pies

2.53

pie

2.16

Pie

2.03

 cake

1.33

 pizza

1.31

 pudding

1.27

 slice

1.13

 pastry

1.05

Activations Density 0.019%

mentions of the word "pie"

references to pie

No Comments

No Known Activations