INDEX

Explanations

phrases referring to an established concept or entity

oai_token-act-pair · gpt-3.5-turbo

references related to knowledge and understanding of existing concepts or realities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Flavoring

-0.73

gdala

-0.71

usher

-0.70

yrus

-0.66

amn

-0.65

ombs

-0.64

omas

-0.63

rush

-0.63

ourke

-0.63

artifacts

-0.62

POSITIVE LOGITS

 unfolded

0.78

 unfolds

0.78

 wont

0.75

zed

0.72

 dictates

0.68

 progressed

0.67

phr

0.65

iHUD

0.64

kj

0.64

 relates

0.63

Activations Density 0.248%

phrases referring to an established concept or entity

references related to knowledge and understanding of existing concepts or realities

No Comments

No Known Activations