INDEX

Explanations

names and related terms associated with the name "Sh" at varying strengths of activation

oai_token-act-pair · gpt-3.5-turbo

repeated references to the character or entity "Sh."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 Leth

-0.78

 commons

-0.72

 Titanic

-0.71

 deficit

-0.69

 numbered

-0.67

 Lans

-0.66

 Ajax

-0.64

 seeing

-0.64

 Legend

-0.63

 coupled

-0.63

POSITIVE LOGITS

akespe

1.33

rimp

1.32

ampoo

1.31

aft

1.28

rapnel

1.27

oppers

1.25

rek

1.25

akedown

1.24

orter

1.23

apeshifter

1.22

Activations Density 0.019%

names and related terms associated with the name "Sh" at varying strengths of activation

repeated references to the character or entity "Sh."

No Comments

No Known Activations