INDEX

Explanations

structured formats or patterns in text, likely indicating document sections or headings

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

. , again Autowired

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

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

 transfieras

-0.85

 مرئيه

-0.83

 мәкал

-0.80

tvguidetime

-0.79

ðsíða

-0.77

 صوتيه

-0.75

Autoritní

-0.73

majánló

-0.72

contentLoaded

-0.72

 ویکی‌پدی

-0.72

POSITIVE LOGITS

0.49

0.45

 Again

0.45

*/

0.44

 again

0.43

↵↵

0.42

Autowired

0.41

Again

0.40

he

0.39

↵

0.39

Activations Density 0.031%

structured formats or patterns in text, likely indicating document sections or headings

. , again Autowired

No Comments

No Known Activations