INDEX

Explanations

language constructs related to identifiers and class definitions in programming contexts

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

creating or naming things

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

expliquer

-0.37

chehen

-0.35

outlook

-0.35

 pogod

-0.34

tidase

-0.33

écrire

-0.32

 kháng

-0.32

 mắt

-0.32

 sekali

-0.31

arrêter

-0.31

POSITIVE LOGITS

BufferException

0.68

 disambiguazione

0.66

 AssemblyProduct

0.57

VersionUID

0.56

Spoljašnje

0.56

CodeAttribute

0.55

]")]

0.53

 дописавши

0.52

fillType

0.51

 الرياضيه

0.51

Activations Density 0.093%

language constructs related to identifiers and class definitions in programming contexts

creating or naming things

No Comments

No Known Activations