INDEX

Explanations

descriptions related to email newsletters and subscription invitations

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Poc

-0.68

 queens

-0.63

 narrator

-0.62

NAT

-0.60

 hypoc

-0.60

 polyg

-0.60

 coord

-0.59

 phil

-0.55

ham

-0.54

 elves

-0.54

POSITIVE LOGITS

Increases

0.72

etry

0.71

 Flavoring

0.69

 Juice

0.69

Trend

0.67

ortment

0.67

 giveaways

0.65

lift

0.64

isure

0.63

hazard

0.62

Activations Density 0.021%

descriptions related to email newsletters and subscription invitations

No Comments

No Known Activations