INDEX

Explanations

references to URLs and diverse numeric or code-like data

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

links to online examples

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

findpost

-0.69

+#+

-0.64

SBATCH

-0.61

GEBURTSDATUM

-0.60

曖昧さ回避

-0.60

 незавершена

-0.59

UrlResolution

-0.59

fjspx

-0.58

 jsPsych

-0.57

المكان

-0.57

POSITIVE LOGITS

almaz

0.33

 aplican

0.32

esanan

0.32

Kết

0.30

empatan

0.30

 conhece

0.30

 Zusammen

0.30

Tempo

0.29

ToJson

0.29

 conocen

0.28

Activations Density 0.053%

references to URLs and diverse numeric or code-like data

links to online examples

No Comments

No Known Activations