INDEX

Explanations

phrases related to negative descriptions or criticism

oai_token-act-pair · gpt-3.5-turbo

descriptors related to negative experiences or qualities

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

RIC

-0.75

riz

-0.74

INTON

-0.69

 scholarship

-0.67

rigan

-0.66

lite

-0.65

rique

-0.64

work

-0.64

leaders

-0.64

rist

-0.63

POSITIVE LOGITS

éĹĺ

0.83

å£«

0.82

 Logic

0.71

dayName

0.70

 Archangel

0.70

 Brach

0.70

 Stress

0.69

 Tyrann

0.68

 Beasts

0.68

 Beast

0.68

Activations Density 0.040%

phrases related to negative descriptions or criticism

descriptors related to negative experiences or qualities

No Comments

No Known Activations