INDEX

Explanations

common conjunctions and quantifying phrases

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

Each succeeding generation

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

DispatchToProps

-0.42

InjectAttribute

-0.40

 Brains

-0.36

mutter

-0.36

mian

-0.36

 Donnie

-0.36

羞

-0.35

 Transition

-0.34

 Shame

-0.34

zional

-0.34

POSITIVE LOGITS

 ब्रेकडाउन

0.74

SourceChecksum

0.65

+#+#

0.60

Personendaten

0.55

évaluateur

0.52

 Italijanski

0.50

 surla

0.50

IsContent

0.50

rungsseite

0.50

AutoField

0.49

Activations Density 0.095%

common conjunctions and quantifying phrases

Each succeeding generation

No Comments

No Known Activations