INDEX

Explanations

mentions of social media platforms and actions related to sharing content

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 protection

-0.70

idates

-0.68

 defects

-0.68

 externalToEVAOnly

-0.66

 dise

-0.65

 contaminants

-0.64

amount

-0.64

lethal

-0.64

 rehe

-0.63

osta

-0.63

POSITIVE LOGITS

 LinkedIn

1.70

 Tumblr

1.67

 Facebook

1.66

 Twitter

1.63

 Instagram

1.63

Twitter

1.61

 Pinterest

1.59

facebook

1.59

 Youtube

1.56

 Reddit

1.55

Activations Density 0.158%

mentions of social media platforms and actions related to sharing content

No Comments

No Known Activations