INDEX

Explanations

phrases related to positive outcomes or beneficial effects

oai_token-act-pair · gpt-3.5-turbo

terms denoting beneficial outcomes or effects

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

 Tall

-1.06

arta

-0.69

strip

-0.67

asers

-0.67

 Stri

-0.66

 GÃ¶

-0.64

rack

-0.63

 Pasadena

-0.63

nai

-0.63

 sacked

-0.60

POSITIVE LOGITS

 beneficial

2.13

 Gillespie

1.42

 therapeutic

1.26

 advantageous

1.21

 Rogue

1.13

 benefic

1.11

 detrimental

1.10

 harmful

1.07

 Benef

1.04

 profitable

1.03

Activations Density 0.022%

phrases related to positive outcomes or beneficial effects

terms denoting beneficial outcomes or effects

No Comments

No Known Activations