INDEX

Explanations

proper names or nouns with 'vo' in them

oai_token-act-pair · gpt-3.5-turbo

references to voices and vocal expressions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

estinal

-0.74

 Alph

-0.65

 Killer

-0.65

ness

-0.61

backer

-0.60

Pub

-0.60

arity

-0.59

stadt

-0.59

 forgiven

-0.58

 Buffett

-0.58

POSITIVE LOGITS

apons

0.91

iture

0.89

yg

0.87

utical

0.87

milo

0.85

hua

0.81

inth

0.79

eer

0.79

illac

0.79

vo

0.77

Activations Density 0.019%

proper names or nouns with 'vo' in them

references to voices and vocal expressions

No Comments

No Known Activations