INDEX

Explanations

phrases related to forming an argument or discussion

oai_token-act-pair · gpt-3.5-turbo

instances of punctuations and connective phrases that indicate ongoing discourse

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

mouth

-0.77

è£ħ

-0.67

adoes

-0.66

ļéĨĴ

-0.65

Catalog

-0.65

bryce

-0.65

mination

-0.64

efe

-0.64

SpaceEngineers

-0.63

natureconservancy

-0.62

POSITIVE LOGITS

 suffice

1.43

let

1.42

 here

1.30

 lets

1.16

 consider

1.08

 imagine

1.05

we

1.01

 please

1.01

here

0.95

 indulge

0.92

Activations Density 0.233%

phrases related to forming an argument or discussion

instances of punctuations and connective phrases that indicate ongoing discourse

No Comments

No Known Activations