INDEX

Explanations

legal or political terms related to conflicts or disputes

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

terms and phrases associated with aggressive political discourse and conflict

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_post

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

 Compton

-0.56

 offic

-0.54

 Sins

-0.51

occ

-0.50

 Downs

-0.50

 Burg

-0.50

 Cous

-0.49

 Celest

-0.48

 tresp

-0.47

 Carpenter

-0.47

POSITIVE LOGITS

ariat

0.90

 memo

0.78

group

0.78

 brigade

0.76

 playbook

0.75

mong

0.74

ometer

0.70

wagon

0.70

 dossier

0.69

 cooker

0.69

Activations Density 0.405%

legal or political terms related to conflicts or disputes

terms and phrases associated with aggressive political discourse and conflict

No Comments

No Known Activations