INDEX

Explanations

phrases related to events or announcements like reports or meetings

oai_token-act-pair · gpt-3.5-turbo

phrases related to conflicts, challenges, or noteworthy events

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

jad

-0.59

 ÂŃ

-0.58

xit

-0.53

ynt

-0.53

pec

-0.52

ttle

-0.52

blers

-0.51

nikov

-0.51

 voic

-0.51

 inco

-0.50

POSITIVE LOGITS

 Featuring

0.62

 Includes

0.60

================================================================

0.57

Adapt

0.55

 Reviews

0.54

 Ages

0.54

 Volume

0.54

 Built

0.52

 Located

0.52

 Edition

0.52

Activations Density 1.191%

phrases related to events or announcements like reports or meetings

phrases related to conflicts, challenges, or noteworthy events

No Comments

No Known Activations