INDEX

Explanations

special characters, HTML code snippets, and technical terms

oai_token-act-pair · gpt-3.5-turbo

references to specific locations, events, or identifiers

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

inki

-0.66

itte

-0.62

isans

-0.60

ioch

-0.59

tarians

-0.58

nick

-0.58

anan

-0.58

odan

-0.58

xual

-0.56

zinski

-0.55

POSITIVE LOGITS

 Trials

0.65

Ĥİ

0.63

 Measures

0.56

Veh

0.54

Production

0.54

 Goods

0.52

 Cure

0.52

 Promotion

0.50

ACT

0.50

 Crim

0.49

Activations Density 1.538%

special characters, HTML code snippets, and technical terms

references to specific locations, events, or identifiers

No Comments

No Known Activations