INDEX

Explanations

future-oriented verbs indicating actions or plans

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

will describe upcoming actions

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

Negative Logits

 EconPapers

-0.63

 Walkover

-0.62

ſelf

-0.61

Descripció

-0.59

tė

-0.58

KommentareTeilen

-0.57

 BrowserModule

-0.57

ſelves

-0.56

awakeFromNib

-0.56

viewDidLoad

-0.55

POSITIVE LOGITS

 today

0.37

 describe

0.35

いきます

0.33

 utilize

0.32

 explore

0.32

 using

0.31

 discuss

0.31

 зараз

0.31

将会

0.31

 upcoming

0.30

Activations Density 0.042%

future-oriented verbs indicating actions or plans

will describe upcoming actions

No Known Activations