INDEX

Explanations

positive actions or attributes

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate permission, opportunity, or flexibility

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

Cth

-0.64

merce

-0.60

ittens

-0.59

ale

-0.57

 Epidem

-0.56

 Pipeline

-0.56

anish

-0.56

 Zimbabwe

-0.56

 seams

-0.56

 Consortium

-0.55

POSITIVE LOGITS

thood

0.83

 choice

0.74

 opportunity

0.70

 chance

0.70

ãĥİ

0.68

license

0.68

ãĥĸ

0.65

ppo

0.61

freedom

0.61

vik

0.60

Activations Density 0.415%

positive actions or attributes

phrases that indicate permission, opportunity, or flexibility

No Comments

No Known Activations