INDEX

Explanations

phrases indicating truth or accuracy

oai_token-act-pair · gpt-3.5-turbo

phrases affirming the concept of truth

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

asers

-0.72

foreseen

-0.72

 Citiz

-0.70

ĪĴ

-0.69

adish

-0.68

antha

-0.68

ADRA

-0.66

flush

-0.65

uled

-0.63

artney

-0.63

POSITIVE LOGITS

hood

0.89

terday

0.75

 believers

0.75

 portrayal

0.68

 enough

0.68

of

0.67

fulness

0.67

0.66

 STORY

0.66

 imprisonment

0.66

Activations Density 0.024%

phrases indicating truth or accuracy

phrases affirming the concept of truth

No Comments

No Known Activations