INDEX

Explanations

text related to formal documentation and reporting

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

approved by

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

 couteau

-0.43

aga

-0.40

 artificiales

-0.39

 doigt

-0.39

 aisladas

-0.38

ViewFeatures

-0.38

 invitamos

-0.38

 couvercle

-0.37

Beli

-0.37

 bundles

-0.36

POSITIVE LOGITS

findpost

0.65

脚注の使い方

0.57

verifyException

0.55

ArrowToggle

0.54

ьаж

0.53

RTLU

0.52



0.49

 autorytatywna

0.49

RegressionTest

0.49

gypti

0.48

Activations Density 0.079%

text related to formal documentation and reporting

approved by

No Comments

No Known Activations