INDEX

Explanations

claimed statements or theories that have been proven false or incorrect

oai_token-act-pair · gpt-3.5-turbo

terms associated with debunking, disproving, or discrediting claims or ideas

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ebus

-0.99

 actionGroup

-0.80

NetMessage

-0.77

 eyed

-0.74

arta

-0.65

aws

-0.65

hold

-0.65

alez

-0.65

son

-0.63

ouver

-0.62

POSITIVE LOGITS

 debunked

1.15

 debunk

1.10

 discredited

0.94

 dispro

0.94

idated

0.92

 unfounded

0.87

ãĤ©

0.85

 refute

0.84

proof

0.82

 refuted

0.80

Activations Density 0.016%

claimed statements or theories that have been proven false or incorrect

terms associated with debunking, disproving, or discrediting claims or ideas

No Comments

No Known Activations