INDEX

Explanations

identifiers and references related to research publications

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

########.

-0.52

stdc

-0.46

 surla

-0.46

-0.45

principalTable

-0.44

TestCase

-0.43

InstanceState

-0.43

 AssemblyTitle

-0.43

ɵɵ

-0.40

IContainer

-0.40

POSITIVE LOGITS

orcid

1.40

ORCID

1.24

 ISNI

0.95

VIAF

0.69

 linkedin

0.58

Autoritní

0.56

asteroide

0.56

Hentet

0.55

 LinkedIn

0.53

 Linkedin

0.53

Activations Density 0.007%

identifiers and references related to research publications

No Comments

No Known Activations