INDEX

Explanations

references to Roman-related terms

oai_token-act-pair · gpt-3.5-turbo

references to Roman history, culture, or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

-1.05

mble

-0.84

NetMessage

-0.80

linger

-0.79

*/(

-0.79

olulu

-0.79

ramid

-0.76

lesh

-0.76

ridges

-0.76

razil

-0.73

POSITIVE LOGITS

 Catholic

1.05

 Catholicism

0.98

 Reign

0.92

 legions

0.88

 numer

0.87

 Catholics

0.84

 Inquisition

0.83

 Torch

0.82

 Emperor

0.82

ians

0.80

Activations Density 0.038%

references to Roman-related terms

references to Roman history, culture, or entities

No Comments

No Known Activations