INDEX

Explanations

specific identity labels based on group affiliation or personal characteristics

oai_token-act-pair · gpt-3.5-turbo

references to identity and social categorization related to ethnicity, sexuality, and other personal attributes

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

attribute

-0.63

ILA

-0.62

SAM

-0.62

Theme

-0.62

Disk

-0.60

Gle

-0.60

prise

-0.60

Random

-0.59

Wo

-0.58

Grant

-0.58

POSITIVE LOGITS

 enough

0.79

blind

0.75

aware

0.75

 aware

0.67

 Enough

0.67

 anyway

0.67

wired

0.66

 asleep

0.66

usional

0.65

enough

0.65

Activations Density 0.391%

specific identity labels based on group affiliation or personal characteristics

references to identity and social categorization related to ethnicity, sexuality, and other personal attributes

No Comments

No Known Activations