INDEX

Explanations

structural HTML elements and programming constructs in the text

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

conditional independence

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

 يتيمه

-0.44

transQ

-0.44

 Chriftian

-0.43



-0.43

onedDateTime

-0.42

 itſelf

-0.42

 fhort

-0.41

Tembelea

-0.39

 dağı

-0.38

Gön

-0.37

POSITIVE LOGITS

mord

1.77

 mord

0.93

 Mord

0.77

Mord

0.71

Personendaten

0.66

بوابة

0.59

ftagPool

0.58

 AssemblyTitle

0.56

ValueStyle

0.56

ParallelGroup

0.56

Activations Density 0.001%

structural HTML elements and programming constructs in the text

conditional independence

No Comments

No Known Activations