INDEX

Explanations

Roman-related terms

oai_token-act-pair · gpt-3.5-turbo

mentions of the Roman Empire and related terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

intosh

-1.02

mble

-1.01

*/(

-0.88

ramid

-0.86

ilater

-0.83

NetMessage

-0.81

jri

-0.80

earance

-0.77

trak

-0.77

efeated

-0.77

POSITIVE LOGITS

 Catholic

1.04

 Roman

0.98

 Torch

0.93

 numer

0.89

 Reign

0.86

 Catholicism

0.84

 Abram

0.81

 Pont

0.81

 Pagan

0.79

 Inquisition

0.79

Activations Density 0.007%

Roman-related terms

mentions of the Roman Empire and related terms

No Comments

No Known Activations