INDEX

Explanations

phrases indicating recent actions or completions

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

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

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

 ulike

-0.30

 eventually

-0.30

évaluateur

-0.29

 дальнейшем

-0.28

 antaranya

-0.28

|)

-0.27

रीद

-0.27

})*/

-0.27

 nantinya

-0.27

 entière

-0.27

POSITIVE LOGITS

recently

0.81

 freshly

0.81

Newly

0.79

 frisch

0.76

 recently

0.76

 recién

0.76

Recently

0.76

 Newly

0.75

newly

0.74

 appena

0.74

Activations Density 0.307%

phrases indicating recent actions or completions

No Comments

No Known Activations