INDEX

Explanations

mentions of people's names and their roles or titles in various situations

oai_token-act-pair · gpt-3.5-turbo

references to various roles within the media and entertainment industry

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ramid

-0.75

 advantageous

-0.69

 rubbish

-0.69

 recognise

-0.69

 handshake

-0.68

 torpedo

-0.68

 recognised

-0.68

 cloning

-0.67

 usefulness

-0.66

 duplication

-0.66

POSITIVE LOGITS

âĢ

2.38

 âĢ

2.03

ãĢ

1.62

âĢł

1.56

âĹı

1.49

ï¸ı

1.48

âľ

1.48

Â»

1.44

 âľ

1.44

Â¶

1.40

Activations Density 0.323%

mentions of people's names and their roles or titles in various situations

references to various roles within the media and entertainment industry

No Comments

No Known Activations