INDEX

Explanations

instances of significant procedural legal terms or references

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

references and notes

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

 isInitialized

-0.51

izante

-0.45

 défendre

-0.45

føl

-0.43

 mijne

-0.43

 TextInputType

-0.43

ckså

-0.42

ücksich

-0.42

 henvis

-0.42

 prijzen

-0.41

POSITIVE LOGITS

脚注の使い方

0.61

XtraEditors

0.52

 виправивши

0.52

 дописавши

0.50

Примечания

0.50

ResumeLayout

0.50

Билгалдахарш

0.47

তথ্যসূত্র

0.45

Referencie

0.45

 drap

0.44

Activations Density 0.019%

instances of significant procedural legal terms or references

references and notes

No Comments

No Known Activations