INDEX

Explanations

positive emotional words or sentiments, particularly focused on love and affection

oai_token-act-pair · gpt-3.5-turbo

references to emotions, particularly related to hearts and feelings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

•layer 5 - love

No Comments

Negative Logits

ression

-0.67

Recomm

-0.66

Delivery

-0.65

ressive

-0.63

ECH

-0.62

ron

-0.62

rav

-0.60

redo

-0.59

Relations

-0.58

 elim

-0.58

POSITIVE LOGITS

chool

1.57

paces

1.49

pace

1.37

mith

1.35

creen

1.33

ystem

1.26

hips

1.26

pring

1.26

ynthesis

1.20

hip

1.19

Activations Density 0.086%

positive emotional words or sentiments, particularly focused on love and affection

references to emotions, particularly related to hearts and feelings

No Comments

No Known Activations