INDEX

Explanations

phrases that emphasize the significance or impact of a specific fact or condition

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

mere fact

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

:✨

-0.40

StoryboardSegue

-0.36

💼

-0.35

 branche

-0.34

AddAttribute

-0.34

 Kedua

-0.34

enumi

-0.33

才是

-0.33

 operators

-0.32

@[+][

-0.32

POSITIVE LOGITS

OGND

0.60

 Allein

0.56

 suffit

0.56

 AssemblyTitle

0.51

itself

0.50

 كومونز

0.50

 Chwiliwch

0.50

だけでも

0.50

الحياه

0.47

 wikipagina

0.47

Activations Density 0.026%

phrases that emphasize the significance or impact of a specific fact or condition

mere fact

No Comments

No Known Activations