INDEX

Explanations

words related to supplementary content or extras in media collections

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

Special Edition commentary

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

rungsseite

-0.52

 disambiguazione

-0.46

 jongen

-0.44

Cyfeiriadau

-0.42

 fondament

-0.42

 lavoratori

-0.41

delwed

-0.40

 cooperación

-0.40

 resourceCulture

-0.39

 flexível

-0.39

POSITIVE LOGITS

OGND

0.57

TagMode

0.50

MLLoader

0.50

 commentary

0.48

formation

0.48

⤹

0.48

 commentaries

0.48

 Reels

0.47

 supervis

0.47

 interviews

0.46

Activations Density 0.022%

words related to supplementary content or extras in media collections

Special Edition commentary

No Comments

No Known Activations