INDEX

Explanations

phrases related to making information or actions publicly available

oai_token-act-pair · gpt-3.5-turbo

phrases related to public disclosure or statements made publicly

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ective

-0.74

gypt

-0.71

 sweeps

-0.69

 endurance

-0.68

aos

-0.68

acements

-0.66

 Maintenance

-0.65

order

-0.63

opal

-0.63

elta

-0.63

POSITIVE LOGITS

 publicly

3.66

 privately

2.33

 openly

1.95

 public

1.58

public

1.57

 formally

1.38

 PUBLIC

1.37

 officially

1.35

 anonymously

1.30

 verbally

1.28

Activations Density 0.010%

phrases related to making information or actions publicly available

phrases related to public disclosure or statements made publicly

No Comments

No Known Activations