INDEX

Explanations

expressions of intention or future actions related to reading and purchasing

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

future intent

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

 قدر

-0.32

 even

-0.31

MathML

-0.30

 Lock

-0.30

inhalt

-0.29

 print

-0.29

 projected

-0.28

 lock

-0.28

 action

-0.28

 throughout

-0.27

POSITIVE LOGITS

醐

0.74

fjspx

0.69

 հղումներ

0.69

onOptions

0.65

+#+#

0.65

WithMany

0.62

 Administrativna

0.61

rbrakk

0.60

лтамалар

0.60

IsContent

0.60

Activations Density 0.012%

expressions of intention or future actions related to reading and purchasing

future intent

No Comments

No Known Activations