INDEX

Explanations

tangible objects related to law and regulations

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Written" and related publication metadata

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

 refuel

-0.72

 MacArthur

-0.70

ĺħ

-0.68

lement

-0.63

 herpes

-0.62

 Leica

-0.62

enegger

-0.61

 Papua

-0.60

virt

-0.59

 antic

-0.59

POSITIVE LOGITS

Written

2.57

 Published

1.61

ICS

1.61

ICK

1.39

 proves

1.05

 Written

0.88

Available

0.88

umbs

0.86

 Hawkins

0.85

alis

0.81

Activations Density 0.042%

tangible objects related to law and regulations

instances of the word "Written" and related publication metadata

No Comments

No Known Activations