INDEX

Explanations

repeatedly used phrases indicating continuation or addition in text

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

also/further relates, desirable, grateful

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

principalColumn

-0.43

 Korn

-0.42

 getContent

-0.38

Koordin

-0.38

kat

-0.37

hach

-0.37

velocityX

-0.37

alad

-0.36

Daf

-0.36

classID

-0.36

POSITIVE LOGITS

ConstraintMaker

0.65

UnusedPrivate

0.62

MessageOf

0.59

AddTagHelper

0.58

AccessorTable

0.57

UrlResolution

0.56

 disambiguazione

0.55

CppMethod

0.53

Personendaten

0.52

 NSCoder

0.51

Activations Density 0.001%

repeatedly used phrases indicating continuation or addition in text

also/further relates, desirable, grateful

No Comments

No Known Activations