INDEX

Explanations

titles or names that are preceded by "Mr." and sometimes other titles as well

oai_token-act-pair · gpt-3.5-turbo

frequencies of the word "Mr." in various contexts

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

anwhile

-0.97

wcs

-0.89

izu

-0.78

ebook

-0.77

 eleph

-0.72

 coincide

-0.70

obyl

-0.70

arily

-0.68

 calendars

-0.67

gypt

-0.66

POSITIVE LOGITS

ItemTracker

0.87

 Olympia

0.80

 Venture

0.73

 Surge

0.72

 Incredible

0.69

 Else

0.68

Jav

0.68

 Stack

0.67

 Hussein

0.67

 Garrison

0.67

Activations Density 0.044%

titles or names that are preceded by "Mr." and sometimes other titles as well

frequencies of the word "Mr." in various contexts

No Comments

No Known Activations