INDEX

Explanations

phrases related to legal and ethical standards

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 كومونز

-0.54

RTEX

-0.42

Kesimpulan

-0.40

 tersangka

-0.40

 NUKAT

-0.40

audiovisuel

-0.40

bootstrapcdn

-0.39

parsedMessage

-0.39

veloper

-0.39

 flexível

-0.39

POSITIVE LOGITS

 behaviour

0.55

 shouldn

0.52

ThroughAttribute

0.51

 behavior

0.51

的行为

0.48

 unacceptable

0.46

不应该

0.46

behaviour

0.46

behavior

0.44

bag

0.42

Activations Density 0.291%

phrases related to legal and ethical standards

No Comments

No Known Activations