INDEX

Explanations

negations and expressions indicating disagreement or doubt

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

the word "Not".

oai_token-act-pair · claude-3-5-haiku-20241022 Triggered by @neilrathi

the word "not" in various contexts.

oai_token-act-pair · gpt-4o Triggered by @neilrathi

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

-------

-0.48

 applicazione

-0.46

 immagine

-0.45

 deleteUser

-0.45

closeConnection

-0.44

 alimentazione

-0.43

 tutela

-0.43

 drapeau

-0.42

 alcuna

-0.42

 illustrazione

-0.42

POSITIVE LOGITS

not

0.70

RectangleBorder

0.67

 utafitiHapana

0.63

 nakalista

0.60

 nahilalakip

0.60

sidian

0.60

లాలు

0.60

 Niet

0.58

 نه

0.58

Not

0.58

Activations Density 0.073%

negations and expressions indicating disagreement or doubt

the word "Not".

the word "not" in various contexts.

No Comments

No Known Activations