INDEX

Explanations

phrases related to continuing or reading additional information

oai_token-act-pair · gpt-3.5-turbo

references to the act of continuing to read

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ĪĴ

-0.87

oult

-0.77

Ã±o

-0.76

»Ĵ

-0.75

estern

-0.72

opard

-0.72

alm

-0.71

ascal

-0.67

Ĭ±

-0.67

oint

-0.64

POSITIVE LOGITS

...]

0.76

 WATCHED

0.72

 Expand

0.71

 toggle

0.71

 comprehension

0.70

 Below

0.69

 aloud

0.67

ahead

0.65

 âĨĴ

0.64

entials

0.62

Activations Density 0.015%

phrases related to continuing or reading additional information

references to the act of continuing to read

No Comments

No Known Activations