INDEX

Explanations

specific techniques or instructions related to careful planning and execution

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

specific classifications

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

Negative Logits

makeText

-0.45

enderror

-0.45

Upon

-0.39

 Kerja

-0.39

なんと

-0.38

 Upon

-0.38

instancetype

-0.38

withOpacity

-0.37

 fits

-0.37

変わる

-0.37

POSITIVE LOGITS

 noDo

0.65

 plutôt

0.60

Chham

0.56

expandindo

0.52

GEBURTSDATUM

0.52

 helst

0.48

 piuttosto

0.48

 мәкал

0.48

principalColumn

0.47

 rather

0.45

Activations Density 0.092%

specific techniques or instructions related to careful planning and execution

specific classifications

No Known Activations