INDEX

Explanations

terms related to conflict, weaponry, and legal matters

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

terms related to conflict and violence

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

VID

-0.60

ggles

-0.58

 Satisf

-0.57

 Thumbnails

-0.57

 Destination

-0.57

 Presence

-0.56

 Swim

-0.55

edIn

-0.53

 Recomm

-0.53

 Response

-0.53

POSITIVE LOGITS

and

0.87

 ranging

0.81

ensitive

0.78

or

0.78

etc

0.78

 such

0.76

hooting

0.75

0.74

creen

0.74

ocial

0.74

Activations Density 0.610%

terms related to conflict, weaponry, and legal matters

terms related to conflict and violence

No Comments

No Known Activations