INDEX

Explanations

phrases that include the word "with" and its associations

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

followed by an

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

Спољашње

-0.48

 types

-0.46

 queſta

-0.42

 handleClick

-0.41

 présidenti

-0.40

gettyimages

-0.40

 africain

-0.39

 Gerechtigkeit

-0.39

Manbalar

-0.39

 racional

-0.39

POSITIVE LOGITS

 simplifié

0.52

rawDesc

0.49

ValueGeneration

0.44

 portico

0.43

 Sequ

0.41

oree

0.41

uthorized

0.41

 offici

0.40

followed

0.39

RetentionPolicy

0.39

Activations Density 0.001%

phrases that include the word "with" and its associations

followed by an

No Comments

No Known Activations