INDEX

Explanations

concepts and terms related to research methodology and findings

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

words introducing sections or lists

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

:✨

-0.88

rungsseite

-0.60

webElementXpaths

-0.56

 nakalista

-0.55

 >=",

-0.54

Datuak

-0.53

 protoimpl

-0.52

ⓧ

-0.52

 Мексичка

-0.50

 Taktlose

-0.49

POSITIVE LOGITS

onAttach

0.36

 charité

0.34

 faune

0.34

 fondateur

0.33

JButton

0.32

CppCodeGen

0.32

AnchorTagHelper

0.32

Jegyzetek

0.32

 reina

0.32

 réservoir

0.32

Activations Density 0.019%

concepts and terms related to research methodology and findings

words introducing sections or lists

No Comments

No Known Activations