INDEX

Explanations

phrases that indicate limitations or negations, particularly focusing on quantities or availability

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

with and without

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

 perſon

-0.51

出版年

-0.51

IconModule

-0.50

 itſelf

-0.47

Personensuche

-0.46

 themſelves

-0.44

 houſe

-0.44

 lendemain

-0.43

bibfield

-0.43

 ſtand

-0.42

POSITIVE LOGITS

with

0.68

With

0.65

 WITH

0.65

 With

0.61

Avec

0.59

 initComponents

0.59

 אית

0.58

WITH

0.57

 Avec

0.56

 עם

0.56

Activations Density 0.043%

phrases that indicate limitations or negations, particularly focusing on quantities or availability

with and without

No Comments

No Known Activations