INDEX

Explanations

different variations of graphical or mathematical representations

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-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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 desierto

-0.47

 antaranya

-0.42

 llorando

-0.40

 orgánico

-0.39

 liitty

-0.38

 infierno

-0.35

 desnuda

-0.35

 vôtre

-0.35

 dalamnya

-0.35

 llorar

-0.34

POSITIVE LOGITS

aarrggbb

0.78

SequentialGroup

0.63

ThroughAttribute

0.61

ACHUSET

0.60

RTGC

0.58

 autorytatywna

0.58

JAKARTA

0.57

 Italijanski

0.57

autoconfigure

0.55

Aholisi

0.54

Activations Density 0.349%

different variations of graphical or mathematical representations

No Comments

No Known Activations