INDEX

Explanations

words related to social groups or communities, specifically the term "clan."

oai_token-act-pair · gpt-3.5-turbo

references to familial or group affiliations, specifically clans and tribes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Kear

-0.72

Jane

-0.72

Ö¼

-0.68

 Seymour

-0.67

piration

-0.67

*/(

-0.64

Accessory

-0.64

 Jane

-0.62

POS

-0.61

 HuffPost

-0.60

POSITIVE LOGITS

 clans

1.37

 clan

1.24

imore

0.97

mates

0.96

 elders

0.85

 chiefs

0.84

members

0.81

ieft

0.78

 dynasty

0.78

ourage

0.76

Activations Density 0.008%

words related to social groups or communities, specifically the term "clan."

references to familial or group affiliations, specifically clans and tribes

No Comments

No Known Activations