INDEX

Explanations

references to additional information or sections in the document

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

below

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

 ſind

-0.56

 שוליים

-0.51

 لينك

-0.49

SequentialGroup

-0.49

RTEX

-0.48

AccessorTable

-0.48

󠁿

-0.47

menuStrip

-0.45

 Allociné

-0.45

CloseOperation

-0.45

POSITIVE LOGITS

endregion

0.42

↹

0.41

Personendaten

0.40

 below

0.38

ględ

0.37

 Meksiko

0.37

Ӕ

0.37

 ↓,

0.36

StoreMessageInfo

0.34

 Italijani

0.34

Activations Density 0.056%

references to additional information or sections in the document

below

No Comments

No Known Activations