INDEX

Explanations

phrases that indicate opinions or advice about improving situations or making choices

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

"it" followed by a descriptive adjective

np_acts-logits-general · gemini-2.0-flash

making it

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

abstractmethod

-0.48

 plazos

-0.47

 calcetines

-0.44

 abbandon

-0.40

 parcial

-0.38

 prazo

-0.37

 contenedores

-0.36

 profesores

-0.35

 agujas

-0.35

 gobern

-0.34

POSITIVE LOGITS

 ujednoznacz

0.75

⟬

0.75

 become

0.69

 becoming

0.66

Vidite

0.65

Becoming

0.65

 make

0.65

 Becoming

0.64

 becomes

0.62

__((

0.60

Activations Density 0.111%

phrases that indicate opinions or advice about improving situations or making choices

"it" followed by a descriptive adjective

making it

No Comments

No Known Activations