INDEX

Explanations

references to programming concepts and structures

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

application level structures

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



-0.36

Dw

-0.32

 autorités

-0.32

viv

-0.31

 papy

-0.31

டைய

-0.31

taminophen

-0.31

 PLOS

-0.30

StringBuilder

-0.30

@[+][

-0.30

POSITIVE LOGITS

GEBURTSDATUM

0.63

 оригіналу

0.59

Personendaten

0.59

 المعيارى

0.58

 <=",

0.54

CloseOperation

0.51

ParallelGroup

0.50

CppMethod

0.49

SequentialGroup

0.46

RootElement

0.45

Activations Density 0.160%

references to programming concepts and structures

application level structures

No Comments

No Known Activations