INDEX

Explanations

terms related to ancient Rome

oai_token-act-pair · gpt-3.5-turbo

references to the Roman context or culture

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

intosh

-1.10

mble

-0.94

ramid

-0.87

*/(

-0.86

olulu

-0.84

jri

-0.83

anwhile

-0.82

NetMessage

-0.82

lessly

-0.82

ickr

-0.81

POSITIVE LOGITS

 Catholic

1.03

 Reign

0.87

 numer

0.87

 Roman

0.86

 Catholicism

0.86

 Torch

0.85

 Catholics

0.81

 Emperor

0.78

 Inquisition

0.78

 Pont

0.77

Activations Density 0.014%

terms related to ancient Rome

references to the Roman context or culture

No Comments

No Known Activations