INDEX

Explanations

proper nouns, which may include brand names, names of companies, and names of media franchises

oai_token-act-pair · gpt-3.5-turbo

names of companies, brands, or organizations

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

enegger

-0.68

.).

-0.60

 è£ı

-0.58

ãĤ¤ãĥĪ

-0.56

orage

-0.55

okemon

-0.54

inez

-0.54

 Niet

-0.53

});

-0.53

 cleaners

-0.52

POSITIVE LOGITS

ipedia

0.65

icer

0.65

moon

0.64

erv

0.62

hou

0.62

Plot

0.60

culus

0.60

usk

0.58

NET

0.57

haw

0.57

Activations Density 0.674%

proper nouns, which may include brand names, names of companies, and names of media franchises

names of companies, brands, or organizations

No Comments

No Known Activations