INDEX

Explanations

words related to URLs and web addresses

oai_token-act-pair · gpt-3.5-turbo

web URLs or links in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 bark

-0.73

Obi

-0.71

 clin

-0.70

 Maher

-0.70

 firing

-0.67

 creep

-0.66

 stiff

-0.65

 Emmy

-0.64

 McCartney

-0.64

 Constantine

-0.63

POSITIVE LOGITS

wp

1.51

forums

1.48

pmwiki

1.47

cgi

1.40

english

1.35

archives

1.34

forum

1.34

uploads

1.29

index

1.28

detail

1.28

Activations Density 0.037%

words related to URLs and web addresses

web URLs or links in the text

No Comments

No Known Activations