INDEX

Explanations

code-related terms and concepts pertaining to programming logic and structures

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

opening page url navigate

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

TypedDataSet

-0.41

ModelSerializer

-0.33

 moeite

-0.33

 JADX

-0.32

vician

-0.32

 Geduld

-0.31

 حی

-0.31

 posiad

-0.30

 sledo

-0.30

 nadzie

-0.29

POSITIVE LOGITS

 typelib

0.56

 disambiguazione

0.56

isition

0.49

minder

0.48

 houſe

0.46

⤹

0.45

tension

0.45

 ſche

0.44

 Quatre

0.44

Efq

0.44

Activations Density 0.020%

code-related terms and concepts pertaining to programming logic and structures

opening page url navigate

No Comments

No Known Activations