INDEX

Explanations

the name "Wu"

oai_token-act-pair · gpt-3.5-turbo

mentions of a specific individual, particularly focused on 'Wu'

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

Scot

-0.69

 simul

-0.68

sed

-0.66

fare

-0.64

 Scot

-0.64

ball

-0.63

 Iraqis

-0.62

abies

-0.62

IND

-0.62

par

-0.62

POSITIVE LOGITS

Wu

4.05

wu

1.64

 Huang

1.63

 Zhou

1.60

Zhu

1.58

Xie

1.52

Liu

1.49

 Gong

1.45

Yao

1.43

Wei

1.42

Activations Density 0.028%

the name "Wu"

mentions of a specific individual, particularly focused on 'Wu'

No Comments

No Known Activations