INDEX

Explanations

instances of behavior that is done in secret or covertly

oai_token-act-pair · gpt-3.5-turbo

references to covert or secretive activities

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

 Applic

-0.69

orth

-0.68

Dur

-0.65

 consistency

-0.64

ity

-0.63

 availability

-0.63

availability

-0.62

pic

-0.61

 Citation

-0.61

 gradient

-0.61

POSITIVE LOGITS

 secretly

3.59

 covert

1.74

 clandestine

1.60

 quietly

1.54

 unconsciously

1.50

 privately

1.48

 silently

1.42

 anonymously

1.40

 secret

1.39

 undercover

1.33

Activations Density 0.017%

instances of behavior that is done in secret or covertly

references to covert or secretive activities

No Comments

No Known Activations