INDEX

Explanations

positive emotional attributes and values

oai_token-act-pair · gpt-3.5-turbo

words and phrases associated with positive values and concepts of ethics

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

pload

-0.80

aceae

-0.77

adobe

-0.66

iov

-0.64

cients

-0.64

iT

-0.63

ortium

-0.63

ãĥĥãĥī

-0.63

iannopoulos

-0.63

oras

-0.63

POSITIVE LOGITS

 alike

1.36

 respectively

1.11

fulness

0.83

 depending

0.82

 amongst

0.80

 amidst

0.78

 thereof

0.76

 characterize

0.70

 among

0.70

 agendas

0.70

Activations Density 0.268%

positive emotional attributes and values

words and phrases associated with positive values and concepts of ethics

No Comments

No Known Activations