INDEX

Explanations

phrases related to opening or closing actions

oai_token-act-pair · gpt-3.5-turbo

phrases related to opening and closing contexts or applications

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

 bearing

-0.69

 mill

-0.69

 losses

-0.67

 plane

-0.67

 seasoning

-0.66

 meteor

-0.66

pun

-0.65

 dread

-0.65

 planes

-0.64

Isa

-0.64

POSITIVE LOGITS

Open

3.56

open

2.16

 Open

1.83

 OPEN

1.70

Opening

1.66

opens

1.65

 open

1.59

Close

1.47

Clear

1.29

Create

1.27

Activations Density 0.011%

phrases related to opening or closing actions

phrases related to opening and closing contexts or applications

No Comments

No Known Activations