INDEX

Explanations

situations where someone is being acknowledged or credited for something they have done

oai_token-act-pair · gpt-3.5-turbo

phrases that express validation or acknowledgment of credit

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

roleum

-0.72

itive

-0.69

uit

-0.68

arching

-0.67

rieving

-0.63

ilege

-0.61

ente

-0.60

vir

-0.59

rete

-0.59

brance

-0.59

POSITIVE LOGITS

 whatsoever

0.71

 approximation

0.70

ĳå£«

0.69

 backups

0.66

Db

0.64

nw

0.63

 accredited

0.58

 aver

0.57

 anyway

0.57

ascal

0.56

Activations Density 0.118%

situations where someone is being acknowledged or credited for something they have done

phrases that express validation or acknowledgment of credit

No Comments

No Known Activations