INDEX

Explanations

references to family-related events or entities

oai_token-act-pair · gpt-3.5-turbo

references to families across various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

msg

-0.74

posts

-0.71

 braking

-0.67

 paraly

-0.67

 emit

-0.66

redd

-0.66

 artific

-0.66

red

-0.66

 neut

-0.65

rush

-0.64

POSITIVE LOGITS

 Family

3.74

Family

2.94

family

2.38

 Families

2.33

 family

2.22

FAM

1.83

 Parents

1.59

 families

1.59

 familial

1.44

Parents

1.43

Activations Density 0.012%

references to family-related events or entities

references to families across various contexts

No Comments

No Known Activations