INDEX

Explanations

terms related to programming capabilities and references

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

Followed by non-English words

np_acts-logits-general · gemini-2.0-flash

foreign language plural articles

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 30-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_30/width_16k/average_l0_120

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.30.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

 quelli

-0.41

vi

-0.31

Vi

-0.30

 secos

-0.28

 którzy

-0.28

who

-0.28

 который

-0.27

IRM

-0.27

 koji

-0.26

 것이다

-0.26

POSITIVE LOGITS

 autorytatywna

0.80

ագրություններ

0.76

 kurios

0.75

 surla

0.65

 koje

0.65

 које

0.64

LookAnd

0.64

 ब्रेकडाउन

0.64

msgTypes

0.63

operatives

0.63

Activations Density 0.041%

terms related to programming capabilities and references

Followed by non-English words

foreign language plural articles

No Comments

No Known Activations