INDEX

Explanations

legal terminology related to rights and applications

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

request to upgrade appeal

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

 AssemblyVersion

-0.35

 supremacy

-0.34

 faut

-0.33

 AttributeSet

-0.33

uanto

-0.33

 véhic

-0.32

 salário

-0.32

Bromoform

-0.32

 darte

-0.31

 nerede

-0.31

POSITIVE LOGITS

 optionally

0.59

 betweenstory

0.58

 poveznice

0.56

Diwedd

0.54

可以选择

0.53

choice

0.52

 Penga

0.50

 canst

0.49

 choice

0.49

tagHelperRunner

0.49

Activations Density 0.037%

legal terminology related to rights and applications

request to upgrade appeal

No Comments

No Known Activations