INDEX

Explanations

keywords related to Roman history and culture

oai_token-act-pair · gpt-3.5-turbo

references to ancient Rome

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

intosh

-1.11

mble

-0.88

*/(

-0.86

olulu

-0.83

cffffcc

-0.82

raints

-0.82

lasses

-0.82

dash

-0.81

irlf

-0.81

dq

-0.80

POSITIVE LOGITS

 Roman

1.25

 Catholic

1.05

Roman

0.99

 Romans

0.97

 Pagan

0.95

 Reign

0.93

 Catholicism

0.89

 Abram

0.89

 Inquisition

0.87

 Catholics

0.85

Activations Density 0.005%

keywords related to Roman history and culture

references to ancient Rome

No Comments

No Known Activations