INDEX

Explanations

phrases related to legal, ethical, philosophical, and societal concepts

oai_token-act-pair · gpt-3.5-turbo

concepts related to arguments and beliefs about morality and self-interest

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 racks

-0.51

 Ahead

-0.48

çīĪ

-0.46

 tops

-0.45

 Boom

-0.44

 Shots

-0.43

 Drops

-0.43

 Shooting

-0.42

 Lights

-0.42

DAQ

-0.41

POSITIVE LOGITS

 epist

0.51

 caus

0.51

ensable

0.51

blance

0.50

 morally

0.50

 psychologically

0.49

ardless

0.49

 subjective

0.49

 causation

0.48

akespe

0.47

Activations Density 4.349%

phrases related to legal, ethical, philosophical, and societal concepts

concepts related to arguments and beliefs about morality and self-interest

No Comments

No Known Activations