INDEX

Explanations

instances of the word "to" in various contexts indicative of goals or purposes

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

achieving goals and desired outcomes

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

 AssemblyCulture

-0.47

uxxxx

-0.44

 Walkover

-0.43

ramientas

-0.40

 estekak

-0.37

 AssemblyCompany

-0.36

 فريبيس

-0.35

ɵɵelementStart

-0.35

Autoritní

-0.34

 (...)

-0.34

POSITIVE LOGITS

 ensure

0.68

 ensuring

0.65

 maintain

0.61

 maintaining

0.59

 increase

0.58

 provide

0.58

 capture

0.57

 increasing

0.56

 sağlay

0.56

 trekken

0.55

Activations Density 0.033%

instances of the word "to" in various contexts indicative of goals or purposes

achieving goals and desired outcomes

No Comments

No Known Activations