INDEX

Explanations

legal terminology and concepts related to courtroom proceedings and evidence

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

has

-0.42

is

-0.41

 cuivre

-0.37

 uzys

-0.37

 Geben

-0.36

 glücklich

-0.36

 peça

-0.36

 Auft

-0.35

 szczę

-0.35

 geométrico

-0.35

POSITIVE LOGITS

 betweenstory

0.90

IntoConstraints

0.73

IUrlHelper

0.71

parsedMessage

0.71

 виправивши

0.71

+#+#

0.68

oredCriteria

0.66

 otomatig

0.65

 pinulongan

0.63

DockStyle

0.62

Activations Density 3.131%

legal terminology and concepts related to courtroom proceedings and evidence

No Comments

No Known Activations