INDEX

Explanations

derogatory terms and phrases related to weakness or inferiority

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 4-gemmascope-mlp-65k

Configuration

google/gemma-scope-2b-pt-mlp/layer_4/width_65k/average_l0_66

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.4.hook_mlp_out

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

帖最后由

-0.56

errHandler

-0.54

ⓧ

-0.53

 estekak

-0.53

entyfik

-0.53

Policies

-0.52

 élevées

-0.52

Batis

-0.52

policies

-0.50

 irré

-0.49

POSITIVE LOGITS

 propOrder

0.66

 meek

0.66

########.

0.63

 helpless

0.57

wim

0.56

 submissive

0.55

TPE

0.53

 suckers

0.52

 cowards

0.51

WebpackPlugin

0.50

Activations Density 0.110%

derogatory terms and phrases related to weakness or inferiority

No Comments

No Known Activations