INDEX

Explanations

names with the initial "F" followed by a high activation number

oai_token-act-pair · gpt-3.5-turbo

instances of the letter "F" as a standalone character or as part of names and titles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Gemini

-0.71

 visually

-0.62

 Sorcerer

-0.61

Pv

-0.57

 Antarctic

-0.56

nav

-0.56

 undercut

-0.56

 Universities

-0.56

 maternity

-0.55

 machine

-0.55

POSITIVE LOGITS

uchs

1.38

ilipp

1.37

ortun

1.26

ergus

1.24

ruits

1.13

udge

1.12

enton

1.10

ultz

1.09

isher

1.08

ogle

1.07

Activations Density 0.034%

names with the initial "F" followed by a high activation number

instances of the letter "F" as a standalone character or as part of names and titles

No Comments

No Known Activations