INDEX

Explanations

varied linguistic elements and concepts in a multilingual or code-related context

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

Fragments of words, often prefixes/suffixes

np_acts-logits-general · gemini-2.0-flash

contralateral, ponovo

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

 workforce

-0.38

mis

-0.37

 andra

-0.36

 Parole

-0.35

 Geor

-0.35

 Protective

-0.34

 speech

-0.33

 Exchange

-0.33

 limb

-0.32

PUL

-0.32

POSITIVE LOGITS

ValueStyle

0.66

migrationBuilder

0.62

RegressionTest

0.61

EDEFAULT

0.59

AnchorStyles

0.59

reportWebVitals

0.57

 étoient

0.55

EndContext

0.55

ThroughAttribute

0.55

Externé

0.54

Activations Density 0.303%

varied linguistic elements and concepts in a multilingual or code-related context

Fragments of words, often prefixes/suffixes

contralateral, ponovo

No Comments

No Known Activations