INDEX

Explanations

references to "original" and related terms within various contexts, such as descriptions, measurements, and functions

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 للمعارف

-0.40

RenderAtEndOf

-0.40

 creciente

-0.37

putes

-0.36

TagMode

-0.36

 cuota

-0.36

awtextra

-0.35

 yemek

-0.35

 bambú

-0.34

//-->

-0.34

POSITIVE LOGITS

 original

1.06

 Original

1.02

original

1.00

Original

0.99

 ORIGINAL

0.95

 originale

0.91

原始

0.90

ORIGINAL

0.90

原

0.87

 oryginal

0.87

Activations Density 0.855%

references to "original" and related terms within various contexts, such as descriptions, measurements, and functions

No Comments

No Known Activations