INDEX

Explanations

terms associated with reductions or decreases in various contexts

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

decrease/reduction

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

 AssemblyTitle

-0.72

 propOrder

-0.64

topRight

-0.62

ArgumentParser

-0.60

 StringTokenizer

-0.60

 onView

-0.58

contentPadding

-0.57

Koordinaten

-0.57

Vla

-0.57

olia

-0.56

POSITIVE LOGITS

 decrease

1.00

 Decrease

0.99

decrease

0.98

Decrease

0.96

 decreased

0.82

 disminuir

0.74

 decreases

0.71

 decreasing

0.70

 Decreased

0.70

 diminuer

0.68

Activations Density 0.022%

terms associated with reductions or decreases in various contexts

decrease/reduction

No Comments

No Known Activations