INDEX

Explanations

variations of the word "Roman"

oai_token-act-pair · gpt-3.5-turbo

references to Rome or related geographical terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

IELD

-0.69

gettable

-0.66

 powdered

-0.65

Ļ½

-0.65

 attendant

-0.64

ĺħ

-0.63

 proximity

-0.60

 eyed

-0.60

 graded

-0.60

 whiff

-0.60

POSITIVE LOGITS

antics

1.19

ulus

1.09

antically

1.08

anes

1.04

ijn

1.02

aji

1.00

antic

0.98

atic

0.97

ano

0.96

agna

0.92

Activations Density 0.035%

variations of the word "Roman"

references to Rome or related geographical terms

No Comments

No Known Activations