INDEX

Explanations

mentions of sacred or revered items or places

oai_token-act-pair · gpt-3.5-turbo

references to things described as "holy"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

llah

-0.82

KER

-0.79

atown

-0.76

kt

-0.73

*/(

-0.73

phrine

-0.73

PsyNetMessage

-0.72

pora

-0.71

yrinth

-0.70

earcher

-0.70

POSITIVE LOGITS

 scripture

0.89

 communion

0.85

 scriptures

0.85

 holy

0.84

gra

0.84

 Grail

0.80

 Scripture

0.80

 rites

0.78

 relics

0.77

ificial

0.77

Activations Density 0.011%

mentions of sacred or revered items or places

references to things described as "holy"

No Comments

No Known Activations