INDEX

Explanations

words related to performers, particularly in the context of entertainment or arts

oai_token-act-pair · gpt-3.5-turbo

references to individuals or groups involved in performance arts

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

Neh

-0.69

 Gree

-0.66

Chapter

-0.64

Nan

-0.62

 Nest

-0.61

Brown

-0.60

Senate

-0.60

Turkey

-0.60

ement

-0.58

gew

-0.58

POSITIVE LOGITS

 performers

3.76

 performer

3.52

 dancers

1.83

 Perform

1.80

 performances

1.70

 performing

1.63

 dancer

1.60

 singers

1.58

 comedians

1.52

performing

1.49

Activations Density 0.013%

words related to performers, particularly in the context of entertainment or arts

references to individuals or groups involved in performance arts

No Comments

No Known Activations