INDEX

Explanations

actions related to physical force or conflict

oai_token-act-pair · gpt-3.5-turbo

actions involving the use or display of weapons

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

Not in Any Lists

No Comments

Negative Logits

ãĥİ

-0.75

uably

-0.72

è¦ļéĨĴ

-0.71

icipated

-0.71

iversary

-0.67

ahime

-0.66

cedented

-0.63

enei

-0.62

 endors

-0.62

POSITIVE LOGITS

 paramedics

0.73

CPR

0.72

 calmly

0.68

 hurried

0.68

 nervously

0.67

 frantically

0.67

 unsuccessfully

0.67

 beside

0.66

 into

0.66

 briefly

0.65

Activations Density 0.315%

actions related to physical force or conflict

actions involving the use or display of weapons

No Comments

No Known Activations