INDEX

Explanations

the term 'otype' or 'otype' followed by a number indicating significance

oai_token-act-pair · gpt-3.5-turbo

terms related to prototypes and typologies

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Policy

-0.69

Nap

-0.69

 thirst

-0.67

 Lawyers

-0.66

Merc

-0.65

deals

-0.64

 Surge

-0.64

 Renew

-0.64

Fla

-0.64

 Saud

-0.63

POSITIVE LOGITS

otype

4.28

otypes

3.62

otyp

2.84

actory

1.46

gen

1.26

 allele

1.11

 phenotype

1.08

ozyg

1.06

agonist

0.95

 prototyp

0.89

Activations Density 0.034%

the term 'otype' or 'otype' followed by a number indicating significance

terms related to prototypes and typologies

No Comments

No Known Activations