INDEX

Explanations

phrases indicating emphasis or persuasion

oai_token-act-pair · gpt-3.5-turbo

expressions of belief or trust

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

imilar

-0.74

 sidel

-0.71

idian

-0.70

advant

-0.64

wich

-0.64

ouk

-0.61

ipment

-0.60

erness

-0.58

erto

-0.58

iri

-0.57

POSITIVE LOGITS

 Yourself

0.65

 hype

0.62

 admit

0.62

zers

0.61

me

0.61

 expr

0.60

 WHEN

0.60

!:

0.59

 deceive

0.59

 Twice

0.58

Activations Density 0.102%

phrases indicating emphasis or persuasion

expressions of belief or trust

No Comments

No Known Activations