INDEX

Explanations

references to concepts, places, or objects considered sacred or holy

oai_token-act-pair · gpt-3.5-turbo

references to sacred or holy concepts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ocre

-0.86

annis

-0.79

yip

-0.71

better

-0.67

ynt

-0.67

ihilation

-0.66

ettlement

-0.66

 Hust

-0.65

uve

-0.65

akeru

-0.65

POSITIVE LOGITS

 cows

1.19

 precincts

0.97

 rites

0.97

gra

0.96

 geometry

0.94

 scripture

0.94

 scriptures

0.93

ness

0.91

 shrine

0.89

cow

0.87

Activations Density 0.051%

references to concepts, places, or objects considered sacred or holy

references to sacred or holy concepts

No Comments

No Known Activations