Los Angeles has seen 15.77 inches of rainfall since the year began. Seattle has only seen around 12.94 inches since Jan. 1. During a more typical year, Seattle would have seen around 16 inches of rain by now, according to data from the National Weather Service (NWS).

WA drought | Los Angeles has seen more rain than Seattle in 2024 | king5.com

Alarmed by the growing number of fires at abandoned properties, Seattle Fire Chief Harold Scoggins is supporting a new strong-armed solution that cracks down on private property owners in favor of public safety.

Officials concerned by surge in abandoned building fires in Seattle

SEATTLE — The Seattle City Council took a step forward on legislation Tuesday that would grant authority to the Seattle Fire Department (SFD) to tear down vacant, derelict buildings. The SFD chief said he is poised to act as soon as the emergency legislation goes into effect in May.

Seattle outlines legislation, timeline to demolish vacant buildings

After 26 days at sea, Race Crew were thrilled and relieved see the Seattle skyline ahead as they motored into Elliott Bay, with loved ones lining the marina walls, homemade flags waving with pride and local craft beers on ice.

Sunny Seattle welcomes Race 10 winners

As noted in the newest West Seattle Junction Association newsletter, Top Pot Doughnuts is returning to West Seattle, three years after they lost their lease and left Alki. When we contacted Top Pot to ask, a spokesperson replied, “The future is very exciting for Top Pot — and we are thrilled ...

West Seattle Blog… | BIZNOTE: Top Pot Doughnuts returning to West Seattle

Casey Martin is a general assignment reporter who has covered everything from political protests to electric scooters. He is almost always out in the field where the news happens. Casey has reported on extremism, houselessness, politics and Seattle’s nightlife. More stories from Casey Martin »

KUOW - Seattle students walk out of class to demand peace in Gaza

In response to the rally, President and CEO of the Jewish Federation of Greater Seattle Solly Kane expressed concern that rallies like this one can create a climate of fear for the students&#x27; Jewish classmates.

Seattle students walk out for pro-Palestinian rally

Downtown Seattle is getting a new arts hub that will support music, fashion, contemporary art... and beyond.

Cannonball! Bumbershoot producers plan big splash with new arts hub in Seattle

FILE - Texas Attorney General Ken Paxton makes a statement at his office, May 26, 2023, in Austin, Texas. Paxton is dropping a request for a Seattle hospital to hand over records regarding gender-affirming treatment potentially given to children from Texas as part of a lawsuit settlement announced ...

Seattle Children's won’t turn over gender-affirming care records - MyNorthwest.com

SEATTLE - The Seattle Police Department is investigating an alleged assault of a cyclist by two University of Washington football players near campus.

UW football players Tybo Rogers, Diesel Gordon accused of assault on Seattle cyclist

Seatti

seatti

Hey Hackers! Have you noticed the "Story Settings" option when submitting your stories? While many of you are familiar with this powerful tool, some might still wonder what it’s all about. We’re here to break it down for you! By the end of this guide, you’ll know exactly how to use HackerNoon’s Story Settings to add the perfect finishing touches to your already amazing stories. The Story Settings feature in your drafts is a post-publication enhancement tool. Everything you add there works to improve your story after it’s published—like boosting its ranking on search engines, attracting more readers, and distributing it to high-traffic pages, among other benefits. Fill out these fields before submitting your draft, and your story will have the best chance to reach a wide audience! Start Writing Your Next Top Story Today How To Use HackerNoon's Story Settings To Your Advantage 1. Upload a Cool Featured Image! A picture (or a GIF) is worth a thousand words.. Featured images are key to catching readers' attention and setting the tone for your story. One common mistake writers make is adding the featured image directly to the draft page. Instead, use the “Story Settings” and select “Click to upload Featured Image” feature to upload your image properly. Taking a moment to pick or create the perfect featured image will make your story more appealing, engaging, and shareable! Speaking of featured images, explore the AI Image Gallery, where we showcase all the AI creations ever made on HackerNoon. 2. Drop a Note to an Editor! You can now reach out to editors directly through your draft’s Story Settings! Write a custom message or select from our curated list of common draft issues for quicker and more efficient communication! While you may not need this feature every time, our curated list of draft issues is incredibly useful when: There’s a delay in reviewing your draft.
Your draft gets rejected by mistake.
You’re experiencing technical issues.
Something is missing, like text, images, or links.
You want to brainstorm ideas or improve your draft. Our editors typically respond within a day or two after you leave a message. 3. Meta description & TL;DR “Meta Descriptions are one of the easiest and most important ways to increase organic traffic to your story. A well-written meta description can both raise your rankings on search engines and also entice readers to click on your story instead of the competing stories on the SERP.”
-Limarc Ambalina, Former Head of Editorial, HackerNoon Your meta description is like an elevator pitch for your story—it’s your moment to grab attention in just a few words! Keep it concise (155-160 characters), include a target keyword, and give readers a compelling preview of what’s inside. Plus, when HackerNoon shares your story on socials, your meta description becomes the caption, making it even more important to get it right - see an example here. Need tips? Learn more about crafting the perfect meta description here. A TL;DR is a brief summary that highlights your story’s key points, helping readers quickly decide if they want to dive in. We recommend keeping it between 300-500 characters. In our newsletters, we feature your story’s image, TL;DR, and a 'Read More' button, so make it compelling—or let our AI generate one for you! 4. Add Tags Tags are the internet's indexing system. They are the foundation of organizing and distributing content on the internet. When you submit a story to HackerNoon, you get not three, not five, but EIGHT tags. Tags work for you, as a writer, in two important ways: Tags help Google, Bing, DuckDuckGo, and friends find your content, serving it to users searching for those terms.
Tags help HackerNoon distribute your content to top trafficked pages, as well as the most subscribed tags via our customizable newsletter, The Tech Brief. Want to learn more? We’ve written a whole story on it! Check it out here. 5. Emoji Credibility Indicators As a technology publishing platform, we work to facilitate value between reader and writer. Our new release of Emoji Credibility Indicators surfaces essential context to the content within every HackerNoon story. This could be something as simple as telling the reader if the story is a news piece or an opinion piece. On the other hand, there are more complicated things we might want to communicate to the reader, such as whether or not the writer has a vested interest in the companies they’re writing about, or if they used AI to write the piece. How the Emoji Credibility Index shows up on your story settings: Learn the meaning of every Emoji Credibility Indicator here. 6. Repost or Original Story? If you’re reposting a story originally published elsewhere, you can still submit it to HackerNoon and leverage our distribution engine. By selecting the “max readership” option, your story benefits from audio story conversion, newsletter distribution, and RSS feed pickups. For quality content republished from personal blogs, we also provide canonical links to help you grow your audience. In addition to a free editorial review and story distribution, you can get your call to action(s) more prominently featured across HackerNoon by updating your profile call to action. Learn more about reposting articles & canonical linking here. That's it! You've added the final touches and are now ready to submit and share your story with the HackerNoon community! Start Writing Your Next Top Story Today Looking for other templates? Use this template to write about a book you’re reading, or explore more templates here. Last Chance to Win Big at These Writing Contests! 🚀 The AI Writing Contest by Bright Data The AI Writing Contest, powered by Bright Data and HackerNoon taps into the global AI revolution. It invites developers, data scientists, researchers, writers, and everyone interested in AI and Large Language Models (LLMs) to share their insights for a chance to win up to $2,500 in prizes. To enter, write a story with the #ai tag using this writing template. Submissions close December 1st, 2024—just 2 days left! 💰 The Bitcoin Writing Contest by Rootstock The Bitcoin Writing Contest, powered by Rootstock and HackerNoon invites thought leaders, writers, talented developers, or anyone passionate about blockchain technology, to showcase your expertise in various Bitcoin-related topics.  We welcome submissions that cover various Bitcoin topics like Runes, ordinals, layer 2 solutions, sidechains, and more. Submit your story using this template for a chance to win. Submissions close tomorrow—just 1 day left! Hey Hackers! Have you noticed the "Story Settings" option when submitting your stories? While many of you are familiar with this powerful tool, some might still wonder what it’s all about. We’re here to break it down for you! By the end of this guide, you’ll know exactly how to use HackerNoon’s Story Settings to add the perfect finishing touches to your already amazing stories. The Story Settings feature in your drafts is a post-publication enhancement tool. Everything you add there works to improve your story after it’s published—like boosting its ranking on search engines, attracting more readers, and distributing it to high-traffic pages, among other benefits. Fill out these fields before submitting your draft, and your story will have the best chance to reach a wide audience! after Start Writing Your Next Top Story Today Start Writing Your Next Top Story Today Start Writing Your Next Top Story Today Start Writing Your Next Top Story Today How To Use HackerNoon's Story Settings To Your Advantage 1. Upload a Cool Featured Image! 1. Upload a Cool Featured Image! A picture (or a GIF) is worth a thousand words.. A picture (or a GIF) is worth a thousand words.. Featured images are key to catching readers' attention and setting the tone for your story. One common mistake writers make is adding the featured image directly to the draft page. Instead, use the “Story Settings” and select “Click to upload Featured Image” feature to upload your image properly. Taking a moment to pick or create the perfect featured image will make your story more appealing, engaging, and shareable! “Click to upload Featured Image” Speaking of featured images, explore the AI Image Gallery, where we showcase all the AI creations ever made on HackerNoon. Speaking of featured images, explore the AI Image Gallery , where we showcase all the AI creations ever made on HackerNoon. AI Image Gallery AI Image Gallery 2. Drop a Note to an Editor! 2. Drop a Note to an Editor! 2. Drop a Note to an Editor! You can now reach out to editors directly through your draft’s Story Settings! Write a custom message or select from our curated list of common draft issues for quicker and more efficient communication! While you may not need this feature every time, our curated list of draft issues is incredibly useful when: There’s a delay in reviewing your draft. Your draft gets rejected by mistake. You’re experiencing technical issues. Something is missing, like text, images, or links. You want to brainstorm ideas or improve your draft. There’s a delay in reviewing your draft. Your draft gets rejected by mistake. You’re experiencing technical issues. Something is missing, like text, images, or links. You want to brainstorm ideas or improve your draft. Our editors typically respond within a day or two after you leave a message. Our editors typically respond within a day or two after you leave a message. 3. Meta description & TL;DR 3. Meta description & TL;DR “Meta Descriptions are one of the easiest and most important ways to increase organic traffic to your story. A well-written meta description can both raise your rankings on search engines and also entice readers to click on your story instead of the competing stories on the SERP.” -Limarc Ambalina, Former Head of Editorial, HackerNoon “Meta Descriptions are one of the easiest and most important ways to increase organic traffic to your story. A well-written meta description can both raise your rankings on search engines and also entice readers to click on your story instead of the competing stories on the SERP.” - Limarc Ambalina , Former Head of Editorial, HackerNoon Limarc Ambalina Your meta description is like an elevator pitch for your story—it’s your moment to grab attention in just a few words! Keep it concise (155-160 characters), include a target keyword, and give readers a compelling preview of what’s inside. Plus, when HackerNoon shares your story on socials, your meta description becomes the caption, making it even more important to get it right - see an example here . here here Need tips? Learn more about crafting the perfect meta description here. Need tips? Learn more about crafting the perfect meta description here . here here A TL;DR is a brief summary that highlights your story’s key points, helping readers quickly decide if they want to dive in. We recommend keeping it between 300-500 characters. In our newsletters, we feature your story’s image, TL;DR, and a 'Read More' button, so make it compelling—or let our AI generate one for you! 4. Add Tags 4. Add Tags Tags are the internet's indexing system. They are the foundation of organizing and distributing content on the internet. When you submit a story to HackerNoon, you get not three, not five, but EIGHT tags. Tags work for you, as a writer, in two important ways: Tags Tags help Google, Bing, DuckDuckGo, and friends find your content, serving it to users searching for those terms. Tags help HackerNoon distribute your content to top trafficked pages, as well as the most subscribed tags via our customizable newsletter, The Tech Brief. Tags help Google, Bing, DuckDuckGo, and friends find your content, serving it to users searching for those terms. Tags help Google , Bing , DuckDuckGo , and friends find your content, Google Bing DuckDuckGo Tags help HackerNoon distribute your content to top trafficked pages, as well as the most subscribed tags via our customizable newsletter, The Tech Brief. Tags help HackerNoon distribute your content to top trafficked pages, Want to learn more? We’ve written a whole story on it! Check it out here. Want to learn more? We’ve written a whole story on it! Check it out here . here 5. Emoji Credibility Indicators 5. Emoji Credibility Indicators As a technology publishing platform, we work to facilitate value between reader and writer. Our new release of Emoji Credibility Indicators surfaces essential context to the content within every HackerNoon story. This could be something as simple as telling the reader if the story is a news piece or an opinion piece. On the other hand, there are more complicated things we might want to communicate to the reader, such as whether or not the writer has a vested interest in the companies they’re writing about, or if they used AI to write the piece. How the Emoji Credibility Index shows up on your story settings: How the Emoji Credibility Index shows up on your story settings: Learn the meaning of every Emoji Credibility Indicator here. Learn the meaning of every Emoji Credibility Indicator here . here 6. Repost or Original Story? 6. Repost or Original Story? If you’re reposting a story originally published elsewhere, you can still submit it to HackerNoon and leverage our distribution engine. By selecting the “max readership” option, your story benefits from audio story conversion, newsletter distribution, and RSS feed pickups. For quality content republished from personal blogs, we also provide canonical links to help you grow your audience. In addition to a free editorial review and story distribution, you can get your call to action(s) more prominently featured across HackerNoon by updating your profile call to action . updating updating your profile call to action your profile call to action Learn more about reposting articles & canonical linking here. Learn more about reposting articles & canonical linking here . here That's it! You've added the final touches and are now ready to submit and share your story with the HackerNoon community! Start Writing Your Next Top Story Today Start Writing Your Next Top Story Today Start Writing Your Next Top Story Today Start Writing Your Next Top Story Today Looking for other templates? Use this template to write about a book you’re reading, or explore more templates here . this template here Last Chance to Win Big at These Writing Contests! Last Chance to Win Big at These Writing Contests! 🚀 The AI Writing Contest by Bright Data The AI Writing Contest by Bright Data The AI Writing Contest by Bright Data The AI Writing Contest , powered by Bright Data and HackerNoon taps into the global AI revolution. It invites developers, data scientists, researchers, writers, and everyone interested in AI and Large Language Models (LLMs) to share their insights for a chance to win up to $2,500 in prizes. To enter, write a story with the #ai tag using this writing template . Submissions close December 1st, 2024 —just 2 days left ! AI Writing Contest #ai this writing template December 1st, 2024 2 days left 💰 The Bitcoin Writing Contest by Rootstock The Bitcoin Writing Contest by Rootstock The Bitcoin Writing Contest by Rootstock The Bitcoin Writing Contest , powered by Rootstock and HackerNoon invites thought leaders, writers, talented developers, or anyone passionate about blockchain technology, to showcase your expertise in various Bitcoin-related topics.  We welcome submissions that cover various Bitcoin topics like Runes, ordinals, layer 2 solutions, sidechains, and more. Submit your story using this template for a chance to win. Submissions close tomorrow—just 1 day left! Bitcoin Writing Contest this template tomorrow—just 1 day left!

Using HackerNoon's Story Settings to Your Advantage: Tips From HackerNoon Editors

استخدام إعدادات القصة في HackerNoon لصالحك: نصائح من محرري HackerNoon

Выкарыстанне налад гісторыі HackerNoon у вашых інтарэсах: парады рэдактараў HackerNoon

Cómo aprovechar al máximo la configuración de historias de HackerNoon: consejos de los editores de HackerNoon

HackerNoonin tarina-asetusten käyttäminen eduksesi: Vinkkejä HackerNoonin toimittajilta

Memanfaatkan Pengaturan Cerita HackerNoon untuk Keuntungan Anda: Kiat dari Editor HackerNoon

HackerNoon のストーリー設定を有効活用する: HackerNoon 編集者からのヒント

Kosalela ba paramètres ya lisolo ya HackerNoon na litomba na yo: Toli ya ba éditeurs ya HackerNoon

ການນໍາໃຊ້ການຕັ້ງຄ່າເລື່ອງຂອງ HackerNoon ເພື່ອປະໂຫຍດຂອງທ່ານ: ຄໍາແນະນໍາຈາກບັນນາທິການ HackerNoon

HackerNoon stāsta iestatījumu izmantošana jūsu priekšrocībām: HackerNoon redaktoru padomi

Sử dụng Cài đặt Câu chuyện của HackerNoon để có lợi cho bạn: Mẹo từ Biên tập viên HackerNoon

Ukusebenzisa iiSetingi zeBali leHackerNoon kwiNgcebiso yakho: Iingcebiso ezivela kubahleli beHackerNoon

充分利用 HackerNoon 的故事设置：来自 HackerNoon 编辑的提示

Here's a practical guide on how to manage Terraform provider configurations for different Yandex Cloud regions using Terragrunt. What you'll need Terraform >= 1.9.7
Terragrunt >= 0.67.16
Yandex Cloud Provider >= 0.129.0 Setup Steps Let's look at how to use Terragrunt to dynamically create provider configs for Yandex Cloud. I'll break this down into digestible pieces: Basic provider setup
First, we'll set up the base Yandex Cloud config in the root terragrunt.hcl. This will automatically generate versions.tf for each module:
locals {
  tf_providers = {
    yandex = ">= 0.129.0"
  }
}

generate "providers_versions" {
  path      = "versions.tf"
  if_exists = "overwrite"
  contents  = <<EOF
terraform {
  required_version = ">= 1.9.7"

  required_providers {
    yandex = {
      source  = "yandex-cloud/yandex"
      version = "${local.tf_providers.yandex}"
    }
  }
}
EOF
}



Region settings
For regions like the newly created KZ region, additional endpoints need to be specified due to the default configuration for the RU region. We can specify them at the project level, for example env.hcl and the providers.tf is generated dynamically for each module:
locals {
  cloud_id         = "SOME_ID"
  folder_id        = "SOME_ID"
  sa_key_file      = "${get_repo_root()}/key.json"
  endpoint         = "api.yandexcloud.kz:443"       # Region-Specific
  storage_endpoint = "storage.yandexcloud.kz"       # Region-Specific
}

generate "providers_configs" {
  path      = "providers.tf"
  if_exists = "overwrite_terragrunt"
  contents  = <<EOF
provider "yandex" {
  service_account_key_file = "${local.sa_key_file}"
  cloud_id                 = "${local.cloud_id}"
  folder_id                = "${local.folder_id}"
  endpoint                 = "${local.endpoint}"
  storage_endpoint         = "${local.storage_endpoint}"
}
EOF
}



Additional providers
If you're working with Kubernetes / Kubectl / Helm in Terraform, you'll need these additional provider configs to manage your cluster. The simplest and most straightforward solution would be to pass cluster_id from a terragrunt dependency into the called module:
dependencies {
  paths = ["path/to/your/mks"]
}

dependency "mks" {
  config_path = "path/to/your/mks"

  mock_outputs_allowed_terraform_commands = ["init", "validate", "plan", "destroy"]
  mock_outputs_merge_strategy_with_state  = "shallow"
  mock_outputs = {
    cluster_id = "cluster_id"
  }
}

terraform {
  source = "path/to/your/module"
}

inputs = {
  cluster_id = dependency.mks.outputs.cluster_id
  . . .
  <OTHER_INPUTS>
  . . .
} Then use data resources in the module to configure providers: variable "cluster_id" {
  type        = string
  default     = null
  description = "Managed Kubernetes Service cluster ID"
}

data "yandex_kubernetes_cluster" "this" {
  cluster_id = var.cluster_id
}

data "yandex_client_config" "this" {}

provider "kubernetes" {
  host                   = data.yandex_kubernetes_cluster.this.master.0.external_v4_endpoint
  cluster_ca_certificate = data.yandex_kubernetes_cluster.this.master.0.cluster_ca_certificate
  token                  = data.yandex_client_config.this.iam_token
}

provider "helm" {
  kubernetes {
    host                   = data.yandex_kubernetes_cluster.this.master.0.external_v4_endpoint
    cluster_ca_certificate = data.yandex_kubernetes_cluster.this.master.0.cluster_ca_certificate
    token                  = data.yandex_client_config.this.iam_token
  }
}

provider "kubectl" {
  host                   = data.yandex_kubernetes_cluster.this.master.0.external_v4_endpoint
  cluster_ca_certificate = data.yandex_kubernetes_cluster.this.master.0.cluster_ca_certificate
  token                  = data.yandex_client_config.this.iam_token
} Notes Using Terragrunt for configuration management:
Terragrunt simplifies configuration management for multiple environments by dynamically generating provider configurations via the generate block in the .hcl files. This setup allows for easy handling of multi-region deployments from a single configuration source.


Setup JSON key for Terragrunt:
To access the Yandex Cloud resources, place the JSON key for the service account in the root directory of your project. Don't forget to add it to .gitignore. Alternatively you can use a static access key.


Configuring the module:
Remember that even if you don't manage the terraform module directly, you can almost always override the configuration using generate when calling the module in terragrunt. Conclusion This setup gives you a clean way to manage Terraform configs across different Yandex Cloud regions. It handles authentication properly and works well whether you're just using basic cloud resources or diving into Kubernetes and Helm deployments. Here's a practical guide on how to manage Terraform provider configurations for different Yandex Cloud regions using Terragrunt. What you'll need Terraform >= 1.9.7 Terragrunt >= 0.67.16 Yandex Cloud Provider >= 0.129.0 Terraform >= 1.9.7 Terragrunt >= 0.67.16 Yandex Cloud Provider >= 0.129.0 Setup Steps Let's look at how to use Terragrunt to dynamically create provider configs for Yandex Cloud. I'll break this down into digestible pieces: Basic provider setup
First, we'll set up the base Yandex Cloud config in the root terragrunt.hcl. This will automatically generate versions.tf for each module:
locals {
  tf_providers = {
    yandex = ">= 0.129.0"
  }
}

generate "providers_versions" {
  path      = "versions.tf"
  if_exists = "overwrite"
  contents  = <<EOF
terraform {
  required_version = ">= 1.9.7"

  required_providers {
    yandex = {
      source  = "yandex-cloud/yandex"
      version = "${local.tf_providers.yandex}"
    }
  }
}
EOF
} Region settings
For regions like the newly created KZ region, additional endpoints need to be specified due to the default configuration for the RU region. We can specify them at the project level, for example env.hcl and the providers.tf is generated dynamically for each module:
locals {
  cloud_id         = "SOME_ID"
  folder_id        = "SOME_ID"
  sa_key_file      = "${get_repo_root()}/key.json"
  endpoint         = "api.yandexcloud.kz:443"       # Region-Specific
  storage_endpoint = "storage.yandexcloud.kz"       # Region-Specific
}

generate "providers_configs" {
  path      = "providers.tf"
  if_exists = "overwrite_terragrunt"
  contents  = <<EOF
provider "yandex" {
  service_account_key_file = "${local.sa_key_file}"
  cloud_id                 = "${local.cloud_id}"
  folder_id                = "${local.folder_id}"
  endpoint                 = "${local.endpoint}"
  storage_endpoint         = "${local.storage_endpoint}"
}
EOF
} Additional providers
If you're working with Kubernetes / Kubectl / Helm in Terraform, you'll need these additional provider configs to manage your cluster. The simplest and most straightforward solution would be to pass cluster_id from a terragrunt dependency into the called module:
dependencies {
  paths = ["path/to/your/mks"]
}

dependency "mks" {
  config_path = "path/to/your/mks"

  mock_outputs_allowed_terraform_commands = ["init", "validate", "plan", "destroy"]
  mock_outputs_merge_strategy_with_state  = "shallow"
  mock_outputs = {
    cluster_id = "cluster_id"
  }
}

terraform {
  source = "path/to/your/module"
}

inputs = {
  cluster_id = dependency.mks.outputs.cluster_id
  . . .
  <OTHER_INPUTS>
  . . .
} Basic provider setup First, we'll set up the base Yandex Cloud config in the root terragrunt.hcl. This will automatically generate versions.tf for each module: locals {
  tf_providers = {
    yandex = ">= 0.129.0"
  }
}

generate "providers_versions" {
  path      = "versions.tf"
  if_exists = "overwrite"
  contents  = <<EOF
terraform {
  required_version = ">= 1.9.7"

  required_providers {
    yandex = {
      source  = "yandex-cloud/yandex"
      version = "${local.tf_providers.yandex}"
    }
  }
}
EOF
} Basic provider setup Basic provider setup First, we'll set up the base Yandex Cloud config in the root terragrunt.hcl . This will automatically generate versions.tf for each module: terragrunt.hcl versions.tf locals {
  tf_providers = {
    yandex = ">= 0.129.0"
  }
}

generate "providers_versions" {
  path      = "versions.tf"
  if_exists = "overwrite"
  contents  = <<EOF
terraform {
  required_version = ">= 1.9.7"

  required_providers {
    yandex = {
      source  = "yandex-cloud/yandex"
      version = "${local.tf_providers.yandex}"
    }
  }
}
EOF
} locals {
  tf_providers = {
    yandex = ">= 0.129.0"
  }
}

generate "providers_versions" {
  path      = "versions.tf"
  if_exists = "overwrite"
  contents  = <<EOF
terraform {
  required_version = ">= 1.9.7"

  required_providers {
    yandex = {
      source  = "yandex-cloud/yandex"
      version = "${local.tf_providers.yandex}"
    }
  }
}
EOF
} Region settings For regions like the newly created KZ region, additional endpoints need to be specified due to the default configuration for the RU region. We can specify them at the project level, for example env.hcl and the providers.tf is generated dynamically for each module: locals {
  cloud_id         = "SOME_ID"
  folder_id        = "SOME_ID"
  sa_key_file      = "${get_repo_root()}/key.json"
  endpoint         = "api.yandexcloud.kz:443"       # Region-Specific
  storage_endpoint = "storage.yandexcloud.kz"       # Region-Specific
}

generate "providers_configs" {
  path      = "providers.tf"
  if_exists = "overwrite_terragrunt"
  contents  = <<EOF
provider "yandex" {
  service_account_key_file = "${local.sa_key_file}"
  cloud_id                 = "${local.cloud_id}"
  folder_id                = "${local.folder_id}"
  endpoint                 = "${local.endpoint}"
  storage_endpoint         = "${local.storage_endpoint}"
}
EOF
} Region settings Region settings For regions like the newly created KZ region, additional endpoints need to be specified due to the default configuration for the RU region. We can specify them at the project level, for example env.hcl and the providers.tf is generated dynamically for each module: KZ endpoints RU env.hcl providers.tf locals {
  cloud_id         = "SOME_ID"
  folder_id        = "SOME_ID"
  sa_key_file      = "${get_repo_root()}/key.json"
  endpoint         = "api.yandexcloud.kz:443"       # Region-Specific
  storage_endpoint = "storage.yandexcloud.kz"       # Region-Specific
}

generate "providers_configs" {
  path      = "providers.tf"
  if_exists = "overwrite_terragrunt"
  contents  = <<EOF
provider "yandex" {
  service_account_key_file = "${local.sa_key_file}"
  cloud_id                 = "${local.cloud_id}"
  folder_id                = "${local.folder_id}"
  endpoint                 = "${local.endpoint}"
  storage_endpoint         = "${local.storage_endpoint}"
}
EOF
} locals {
  cloud_id         = "SOME_ID"
  folder_id        = "SOME_ID"
  sa_key_file      = "${get_repo_root()}/key.json"
  endpoint         = "api.yandexcloud.kz:443"       # Region-Specific
  storage_endpoint = "storage.yandexcloud.kz"       # Region-Specific
}

generate "providers_configs" {
  path      = "providers.tf"
  if_exists = "overwrite_terragrunt"
  contents  = <<EOF
provider "yandex" {
  service_account_key_file = "${local.sa_key_file}"
  cloud_id                 = "${local.cloud_id}"
  folder_id                = "${local.folder_id}"
  endpoint                 = "${local.endpoint}"
  storage_endpoint         = "${local.storage_endpoint}"
}
EOF
} Additional providers If you're working with Kubernetes / Kubectl / Helm in Terraform, you'll need these additional provider configs to manage your cluster. The simplest and most straightforward solution would be to pass cluster_id from a terragrunt dependency into the called module: dependencies {
  paths = ["path/to/your/mks"]
}

dependency "mks" {
  config_path = "path/to/your/mks"

  mock_outputs_allowed_terraform_commands = ["init", "validate", "plan", "destroy"]
  mock_outputs_merge_strategy_with_state  = "shallow"
  mock_outputs = {
    cluster_id = "cluster_id"
  }
}

terraform {
  source = "path/to/your/module"
}

inputs = {
  cluster_id = dependency.mks.outputs.cluster_id
  . . .
  <OTHER_INPUTS>
  . . .
} Additional providers Additional providers If you're working with Kubernetes / Kubectl / Helm in Terraform , you'll need these additional provider configs to manage your cluster. The simplest and most straightforward solution would be to pass cluster_id from a terragrunt dependency into the called module: Kubernetes Kubectl Helm Terraform cluster_id dependencies {
  paths = ["path/to/your/mks"]
}

dependency "mks" {
  config_path = "path/to/your/mks"

  mock_outputs_allowed_terraform_commands = ["init", "validate", "plan", "destroy"]
  mock_outputs_merge_strategy_with_state  = "shallow"
  mock_outputs = {
    cluster_id = "cluster_id"
  }
}

terraform {
  source = "path/to/your/module"
}

inputs = {
  cluster_id = dependency.mks.outputs.cluster_id
  . . .
  <OTHER_INPUTS>
  . . .
} dependencies {
  paths = ["path/to/your/mks"]
}

dependency "mks" {
  config_path = "path/to/your/mks"

  mock_outputs_allowed_terraform_commands = ["init", "validate", "plan", "destroy"]
  mock_outputs_merge_strategy_with_state  = "shallow"
  mock_outputs = {
    cluster_id = "cluster_id"
  }
}

terraform {
  source = "path/to/your/module"
}

inputs = {
  cluster_id = dependency.mks.outputs.cluster_id
  . . .
  <OTHER_INPUTS>
  . . .
} Then use data resources in the module to configure providers: data variable "cluster_id" {
  type        = string
  default     = null
  description = "Managed Kubernetes Service cluster ID"
}

data "yandex_kubernetes_cluster" "this" {
  cluster_id = var.cluster_id
}

data "yandex_client_config" "this" {}

provider "kubernetes" {
  host                   = data.yandex_kubernetes_cluster.this.master.0.external_v4_endpoint
  cluster_ca_certificate = data.yandex_kubernetes_cluster.this.master.0.cluster_ca_certificate
  token                  = data.yandex_client_config.this.iam_token
}

provider "helm" {
  kubernetes {
    host                   = data.yandex_kubernetes_cluster.this.master.0.external_v4_endpoint
    cluster_ca_certificate = data.yandex_kubernetes_cluster.this.master.0.cluster_ca_certificate
    token                  = data.yandex_client_config.this.iam_token
  }
}

provider "kubectl" {
  host                   = data.yandex_kubernetes_cluster.this.master.0.external_v4_endpoint
  cluster_ca_certificate = data.yandex_kubernetes_cluster.this.master.0.cluster_ca_certificate
  token                  = data.yandex_client_config.this.iam_token
} variable "cluster_id" {
  type        = string
  default     = null
  description = "Managed Kubernetes Service cluster ID"
}

data "yandex_kubernetes_cluster" "this" {
  cluster_id = var.cluster_id
}

data "yandex_client_config" "this" {}

provider "kubernetes" {
  host                   = data.yandex_kubernetes_cluster.this.master.0.external_v4_endpoint
  cluster_ca_certificate = data.yandex_kubernetes_cluster.this.master.0.cluster_ca_certificate
  token                  = data.yandex_client_config.this.iam_token
}

provider "helm" {
  kubernetes {
    host                   = data.yandex_kubernetes_cluster.this.master.0.external_v4_endpoint
    cluster_ca_certificate = data.yandex_kubernetes_cluster.this.master.0.cluster_ca_certificate
    token                  = data.yandex_client_config.this.iam_token
  }
}

provider "kubectl" {
  host                   = data.yandex_kubernetes_cluster.this.master.0.external_v4_endpoint
  cluster_ca_certificate = data.yandex_kubernetes_cluster.this.master.0.cluster_ca_certificate
  token                  = data.yandex_client_config.this.iam_token
} Notes Using Terragrunt for configuration management:
Terragrunt simplifies configuration management for multiple environments by dynamically generating provider configurations via the generate block in the .hcl files. This setup allows for easy handling of multi-region deployments from a single configuration source. Setup JSON key for Terragrunt:
To access the Yandex Cloud resources, place the JSON key for the service account in the root directory of your project. Don't forget to add it to .gitignore. Alternatively you can use a static access key. Configuring the module:
Remember that even if you don't manage the terraform module directly, you can almost always override the configuration using generate when calling the module in terragrunt. Using Terragrunt for configuration management: Terragrunt simplifies configuration management for multiple environments by dynamically generating provider configurations via the generate block in the .hcl files. This setup allows for easy handling of multi-region deployments from a single configuration source. Using Terragrunt for configuration management: Using Terragrunt for configuration management: Terragrunt simplifies configuration management for multiple environments by dynamically generating provider configurations via the generate block in the .hcl files. This setup allows for easy handling of multi-region deployments from a single configuration source. Terragrunt .hcl Setup JSON key for Terragrunt: To access the Yandex Cloud resources, place the JSON key for the service account in the root directory of your project. Don't forget to add it to .gitignore. Alternatively you can use a static access key. Setup JSON key for Terragrunt: Setup JSON key for Terragrunt: To access the Yandex Cloud resources, place the JSON key for the service account in the root directory of your project. Don't forget to add it to .gitignore . Alternatively you can use a static access key . JSON key .gitignore static access key Configuring the module: Remember that even if you don't manage the terraform module directly, you can almost always override the configuration using generate when calling the module in terragrunt. Configuring the module: Configuring the module: Remember that even if you don't manage the terraform module directly, you can almost always override the configuration using generate when calling the module in terragrunt. generate Conclusion This setup gives you a clean way to manage Terraform configs across different Yandex Cloud regions. It handles authentication properly and works well whether you're just using basic cloud resources or diving into Kubernetes and Helm deployments.

Setting up Yandex Cloud Provider with Terraform and Terragrunt

Setting Up Prometheus Alertmanager on GPUs for Improved ML Lifecycle

Configuración de Prometheus Alertmanager en GPU para mejorar el ciclo de vida del aprendizaje automático

ML ライフサイクルの改善のために GPU 上に Prometheus Alertmanager を設定する

ການຕັ້ງຄ່າ Prometheus Alertmanager ໃນ GPUs ສໍາລັບການປັບປຸງ ML Lifecycle

Поставување на Prometheus Alertmanager на графички процесори за подобрен животен циклус на ML

Prometheus Alertmanager op GPU's instellen voor een verbeterde ML-levenscyclus

Prometheus Alertmanager GPUs kaqpi allinchasqa ML Kawsay Ciclo kaqpaq churay

වැඩිදියුණු කළ ML ජීවන චක්‍රය සඳහා GPU මත Prometheus Alertmanager පිහිටුවීම

Konfigurera Prometheus Alertmanager på GPU:er för förbättrad ML-livscykel

Насб кардани Prometheus Alertmanager дар GPUҳо барои такмили давраи ML

ንዝተመሓየሸ ዑደት ህይወት ML ኣብ GPUs Prometheus Alertmanager ምድላው

Gowulandyrylan ML ömri üçin GPU-larda Prometheus Alertmanager gurmak

Ku Veka Prometheus Alertmanager eka ti-GPU ta Xirhendzevutani xa Vutomi bya ML lexi Antswiweke

(1) Feng Liang, The University of Texas at Austin and Work partially done during an internship at Meta GenAI (Email: jeffliang@utexas.edu);
(2) Bichen Wu, Meta GenAI and Corresponding author;
(3) Jialiang Wang, Meta GenAI;
(4) Licheng Yu, Meta GenAI;
(5) Kunpeng Li, Meta GenAI;
(6) Yinan Zhao, Meta GenAI;
(7) Ishan Misra, Meta GenAI;
(8) Jia-Bin Huang, Meta GenAI;
(9) Peizhao Zhang, Meta GenAI (Email: stzpz@meta.com);
(10) Peter Vajda, Meta GenAI (Email: vajdap@meta.com);
(11) Diana Marculescu, The University of Texas at Austin (Email: dianam@utexas.edu). Table of Links Abstract and Introduction
2. Related Work
3. Preliminary
4. FlowVid
4.1. Inflating image U-Net to accommodate video
4.2. Training with joint spatial-temporal conditions
4.3. Generation: edit the first frame then propagate


Experiments


5.1. Settings
5.2. Qualitative results
5.3. Quantitative results
5.4. Ablation study and 5.5. Limitations
Conclusion, Acknowledgments and References
A. Webpage Demo and B. Quantitative comparisons 5. Experiments 5.1. Settings Implementation Details We train our model with 100k videos from Shutterstock [1]. For each training video, we sequentially sample 16 frames with interval {2,4,8}, which represent videos lasting {1,2,4} seconds (taking videos with FPS of 30). The resolution of all images, including input frames, spatial condition images, and flow warped frames, is set to 512×512 via center crop. We train the model with a batch size of 1 per GPU and a total batch size of 8 with 8 GPUs. We employ AdamW optimizer [28] with a learning rate of 1e-5 for 100k iterations. As detailed in our method, we train the major U-Net and ControlNet U-Net joint branches with v-parameterization [41]. The training takes four days on one 8-A100-80G node. During generation, we first generate keyframes with our trained model and then use an off-the-shelf frame interpolation model, such as RIFE [21], to generate non-key frames. By default, we produce 16 key frames at an interval of 4, corresponding to a 2-second clip at 8 FPS. Then, we use RIFE to interpolate the results to 32 FPS. We employ classifier-free guidance [15] with a scale of 7.5 and use 20 inference sampling steps. Additionally, the Zero SNR noise scheduler [27] is utilized. We also fuse the self-attention features obtained during the DDIM inversion of corresponding key frames from the input video, following FateZero [35]. We evaluate our FlowVid with two different spatial conditions: canny edge maps [5] and depth maps [38]. A comparison of these controls can be found in Section 5.4. Evaluation We select the 25 object-centric videos from the public DAVIS dataset [34], covering humans, animals, etc We manually design 115 prompts for these videos, spanning from stylization to object swap. Besides, we also collect 50 Shutterstock videos [1] with 200 designed prompts. We conduct both qualitative (see Section 5.2) and quantitative comparisons (see Section 5.3) with state-of-the-art methods including Rerender [49], CoDeF [32] and TokenFlow [13]. We use their official codes with the default settings. This paper is available on arxiv under CC 4.0 license. (1) Feng Liang, The University of Texas at Austin and Work partially done during an internship at Meta GenAI (Email: jeffliang@utexas.edu); (2) Bichen Wu, Meta GenAI and Corresponding author; (3) Jialiang Wang, Meta GenAI; (4) Licheng Yu, Meta GenAI; (5) Kunpeng Li, Meta GenAI; (6) Yinan Zhao, Meta GenAI; (7) Ishan Misra, Meta GenAI; (8) Jia-Bin Huang, Meta GenAI; (9) Peizhao Zhang, Meta GenAI (Email: stzpz@meta.com); (10) Peter Vajda, Meta GenAI (Email: vajdap@meta.com); (11) Diana Marculescu, The University of Texas at Austin (Email: dianam@utexas.edu). (1) Feng Liang, The University of Texas at Austin and Work partially done during an internship at Meta GenAI (Email: jeffliang@utexas.edu); (2) Bichen Wu, Meta GenAI and Corresponding author; (3) Jialiang Wang, Meta GenAI; (4) Licheng Yu, Meta GenAI; (5) Kunpeng Li, Meta GenAI; (6) Yinan Zhao, Meta GenAI; (7) Ishan Misra, Meta GenAI; (8) Jia-Bin Huang, Meta GenAI; (9) Peizhao Zhang, Meta GenAI (Email: stzpz@meta.com); (10) Peter Vajda, Meta GenAI (Email: vajdap@meta.com); (11) Diana Marculescu, The University of Texas at Austin (Email: dianam@utexas.edu). Table of Links Abstract and Introduction 2. Related Work 3. Preliminary 4. FlowVid 4.1. Inflating image U-Net to accommodate video 4.2. Training with joint spatial-temporal conditions 4.3. Generation: edit the first frame then propagate Experiments 5.1. Settings 5.2. Qualitative results 5.3. Quantitative results 5.4. Ablation study and 5.5. Limitations Conclusion, Acknowledgments and References A. Webpage Demo and B. Quantitative comparisons Abstract and Introduction Abstract and Introduction 2. Related Work 2. Related Work 3. Preliminary 3. Preliminary 4. FlowVid 4. FlowVid 4 .1. Inflating image U-Net to accommodate video .1. Inflating image U-Net to accommodate video 4.2. Training with joint spatial-temporal conditions 4.2. Training with joint spatial-temporal conditions 4.3. Generation: edit the first frame then propagate 4.3. Generation: edit the first frame then propagate Experiments Experiments Experiments 5.1. Settings 5.1. Settings 5.2. Qualitative results 5.2. Qualitative results 5.3. Quantitative results 5.3. Quantitative results 5.4. Ablation study and 5.5. Limitations 5.4. Ablation study and 5.5. Limitations Conclusion, Acknowledgments and References Conclusion, Acknowledgments and References A. Webpage Demo and B. Quantitative comparisons A. Webpage Demo and B. Quantitative comparisons 5. Experiments 5.1. Settings Implementation Details We train our model with 100k videos from Shutterstock [1]. For each training video, we sequentially sample 16 frames with interval {2,4,8}, which represent videos lasting {1,2,4} seconds (taking videos with FPS of 30). The resolution of all images, including input frames, spatial condition images, and flow warped frames, is set to 512×512 via center crop. We train the model with a batch size of 1 per GPU and a total batch size of 8 with 8 GPUs. We employ AdamW optimizer [28] with a learning rate of 1e-5 for 100k iterations. As detailed in our method, we train the major U-Net and ControlNet U-Net joint branches with v-parameterization [41]. The training takes four days on one 8-A100-80G node. Implementation Details During generation, we first generate keyframes with our trained model and then use an off-the-shelf frame interpolation model, such as RIFE [21], to generate non-key frames. By default, we produce 16 key frames at an interval of 4, corresponding to a 2-second clip at 8 FPS. Then, we use RIFE to interpolate the results to 32 FPS. We employ classifier-free guidance [15] with a scale of 7.5 and use 20 inference sampling steps. Additionally, the Zero SNR noise scheduler [27] is utilized. We also fuse the self-attention features obtained during the DDIM inversion of corresponding key frames from the input video, following FateZero [35]. We evaluate our FlowVid with two different spatial conditions: canny edge maps [5] and depth maps [38]. A comparison of these controls can be found in Section 5.4. Evaluation We select the 25 object-centric videos from the public DAVIS dataset [34], covering humans, animals, etc We manually design 115 prompts for these videos, spanning from stylization to object swap. Besides, we also collect 50 Shutterstock videos [1] with 200 designed prompts. We conduct both qualitative (see Section 5.2) and quantitative comparisons (see Section 5.3) with state-of-the-art methods including Rerender [49], CoDeF [32] and TokenFlow [13]. We use their official codes with the default settings. Evaluation This paper is available on arxiv under CC 4.0 license. This paper is available on arxiv under CC 4.0 license. available on arxiv

FlowVid: Taming Imperfect Optical Flows for Consistent Video-to-Video Synthesis: Settings

Authors:
(1) Amador Durán, SCORE Lab, I3US Institute, Universidad de Sevilla, Sevilla, Spain (amador@us.es);
(2) Pablo Fernández, SCORE Lab, I3US Institute, Universidad de Sevilla, Sevilla, Spain (pablofm@us.es);
(3) Beatriz Bernárdez, I3US Institute, Universidad de Sevilla, Sevilla, Spain (beat@us.es);
(4) Nathaniel Weinman, Computer Science Division, University of California, Berkeley, Berkeley, CA, USA (nweinman@berkeley.edu);
(5) Aslı Akalın, Computer Science Division, University of California, Berkeley, Berkeley, CA, USA (asliakalin@berkeley.edu);
(6) Armando Fox, Computer Science Division, University of California, Berkeley, Berkeley, CA, USA (fox@berkeley.edu). Table of Links Abstract and 1 Introduction 1.1 The twincode platform 1.2 Related Work 2 Research Questions 3 Variables 3.1 Independent Variables 3.2 Dependent Variables 3.3 Confounding Variables 4 Participants 5 Execution Plan and 5.1 Recruitment 5.2 Training and 5.3 Experiment Execution 5.4 Data Analysis Acknowledgments and References ABSTRACT Context. Pair programming has been found to increase student interest in Computer Science, particularly so for women, and would therefore appear to be a way to help remedy the under–representation of women in the field. However, one reason for this under– representation is the unwelcoming climate created by gender stereotypes applied to engineers in general, and to software engineers in particular, assuming that men perform better than their women peers. If this same bias is present in pair programming, it could work against the goal of improving gender balance in computing. Objective. In a remote setting in which students cannot directly observe the gender of their peers, we aim to explore whether Software Engineering students behave differently when the perceived gender of their remote pair programming partners changes, searching for differences in (i) the perceived productivity compared to solo programming; (ii) the partner’s perceived technical competency compared to their own; (iii) the partner’s perceived skill level; (iv) the interaction behavior, such as the frequency of source code additions, deletions, validations, etc.; and (v) the type and relative frequencies of dialog messages used for collaborative behavior in a chat window. Although there are some studies on pair programming performance and gender pair combination, to the best of our knowledge there are no studies on the impact of gender stereotypes and bias within the pairs themselves. Method. We have developed an online platform (twincode) that randomly classifies students into gender–balanced groups, arranges them in pairs for remote pair programming (sharing an editor window and a chat window), and can selectively deceive one or both partners regarding the gender of the other via the use of a clearly gendered avatar. Several behaviors are automatically measured during the pair programming process, together with two questionnaires and a semantic tagging of the pairs’ conversations. We will perform a series of experiments to identify the effect, if any, of possible gender bias in remote pair programming interactions. Students in the control group will have no information about their partner’s gender; students in the treatment group will receive such information but will be selectively deceived about their partner’s true gender. To analyze the data, apart from checking reliability of questionnaire data using Cronbach’s alpha and Kaiser criterion, for each response variable we will (i) compare control and experimental groups for the score distance between two in–pair tasks; then, using the data from the experimental group only, we will (ii) compare scores using the partner’s perceived gender as a within–subjects variable; and (iii) analyze the interaction between the partner’s perceived gender (within–subjects) and the subject’s gender (between–subjects). For the (i) and (ii) analyses we will use t–tests, whereas for the (iii) analyses we will use mixed–model ANOVAs. 1 INTRODUCTION Pair programming is an increasingly popular collaboration paradigm that has been shown to be an effective tool in Computer Science education as measured by positive influence on grades, class performance, confidence, productivity, and motivation to stay [6], especially for women [19, 24]. In pair programming, two partners work closely together to solve a programming task. As such, their ability to engage with each other is key. However, these interactions are influenced by implicit gender bias [12, 18], such as assuming women are less technically competent [18]. This is a widely observed phenomenon even in highly–structured settings [6, 13]. Social sciences research indicates that one’s behavior of an individual is affected by the behavior of their peers [8]. Therefore, implicit gender bias based on perception of peers may have effects on one’s behavior, potentially influencing pair programming experience. In this work, in a non–colocated (i.e. remote) environment in which the gender of the peers cannot be directly observed, our goal is to explore whether Software Engineering students change their behavior when the perceived gender of their remote pair programming partners changes from man to woman or vice versa. Note that, while we recognize that many students may identify as neither men nor women, our initial exploration focuses primarily on interactions between students who identify as one of these, so that we can better align our findings with the existing literature on implicit gender bias. The potential biases in interactions involving gender-fluid, non-gender-conforming, or nonbinary students is a rich and complex topic deserving its own subsequent study. To achieve our goal, we plan to search for differences not only in the perceived productivity of pair programming compared to solo programming, the partner’s perceived technical competency compared to their own, and the partners’ perceived skill level, but also in the interaction behavior, i.e. the frequency of source code additions, deletions, validations, etc., and the type and relative frequency of dialog messages used for collaborative behavior. To get early feedback on the infrastructure supporting our proposal, we ran two pilot studies, one at each university, with a limited number of students, where we could check the comprehensibility of the questionnaires used to gather subjective data, the applicability of the message tagging (described in Section 2), and the capabilities of the twincode platform, which is briefly described below. This paper is available on arxiv under CC BY 4.0 DEED license. Authors: (1) Amador Durán, SCORE Lab, I3US Institute, Universidad de Sevilla, Sevilla, Spain (amador@us.es); (2) Pablo Fernández, SCORE Lab, I3US Institute, Universidad de Sevilla, Sevilla, Spain (pablofm@us.es); (3) Beatriz Bernárdez, I3US Institute, Universidad de Sevilla, Sevilla, Spain (beat@us.es); (4) Nathaniel Weinman, Computer Science Division, University of California, Berkeley, Berkeley, CA, USA (nweinman@berkeley.edu); (5) Aslı Akalın, Computer Science Division, University of California, Berkeley, Berkeley, CA, USA (asliakalin@berkeley.edu); (6) Armando Fox, Computer Science Division, University of California, Berkeley, Berkeley, CA, USA (fox@berkeley.edu). Authors: Authors: (1) Amador Durán, SCORE Lab, I3US Institute, Universidad de Sevilla, Sevilla, Spain (amador@us.es); (2) Pablo Fernández, SCORE Lab, I3US Institute, Universidad de Sevilla, Sevilla, Spain (pablofm@us.es); (3) Beatriz Bernárdez, I3US Institute, Universidad de Sevilla, Sevilla, Spain (beat@us.es); (4) Nathaniel Weinman, Computer Science Division, University of California, Berkeley, Berkeley, CA, USA (nweinman@berkeley.edu); (5) Aslı Akalın, Computer Science Division, University of California, Berkeley, Berkeley, CA, USA (asliakalin@berkeley.edu); (6) Armando Fox, Computer Science Division, University of California, Berkeley, Berkeley, CA, USA (fox@berkeley.edu). Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 1.1 The twincode platform 1.1 The twincode platform 1.2 Related Work 1.2 Related Work 2 Research Questions 2 Research Questions 3 Variables 3 Variables 3.1 Independent Variables 3.1 Independent Variables 3.2 Dependent Variables 3.2 Dependent Variables 3.3 Confounding Variables 3.3 Confounding Variables 4 Participants 4 Participants 5 Execution Plan and 5.1 Recruitment 5 Execution Plan and 5.1 Recruitment 5.2 Training and 5.3 Experiment Execution 5.2 Training and 5.3 Experiment Execution 5.4 Data Analysis 5.4 Data Analysis Acknowledgments and References Acknowledgments and References ABSTRACT Context . Pair programming has been found to increase student interest in Computer Science, particularly so for women, and would therefore appear to be a way to help remedy the under–representation of women in the field. However, one reason for this under– representation is the unwelcoming climate created by gender stereotypes applied to engineers in general, and to software engineers in particular, assuming that men perform better than their women peers. If this same bias is present in pair programming, it could work against the goal of improving gender balance in computing. Objective. In a remote setting in which students cannot directly observe the gender of their peers, we aim to explore whether Software Engineering students behave differently when the perceived gender of their remote pair programming partners changes, searching for differences in (i) the perceived productivity compared to solo programming; (ii) the partner’s perceived technical competency compared to their own; (iii) the partner’s perceived skill level; (iv) the interaction behavior, such as the frequency of source code additions, deletions, validations, etc.; and (v) the type and relative frequencies of dialog messages used for collaborative behavior in a chat window. Although there are some studies on pair programming performance and gender pair combination, to the best of our knowledge there are no studies on the impact of gender stereotypes and bias within the pairs themselves. Method. We have developed an online platform (twincode) that randomly classifies students into gender–balanced groups, arranges them in pairs for remote pair programming (sharing an editor window and a chat window), and can selectively deceive one or both partners regarding the gender of the other via the use of a clearly gendered avatar. Several behaviors are automatically measured during the pair programming process, together with two questionnaires and a semantic tagging of the pairs’ conversations. We will perform a series of experiments to identify the effect, if any, of possible gender bias in remote pair programming interactions. Students in the control group will have no information about their partner’s gender; students in the treatment group will receive such information but will be selectively deceived about their partner’s true gender. To analyze the data, apart from checking reliability of questionnaire data using Cronbach’s alpha and Kaiser criterion, for each response variable we will (i) compare control and experimental groups for the score distance between two in–pair tasks; then, using the data from the experimental group only, we will (ii) compare scores using the partner’s perceived gender as a within–subjects variable; and (iii) analyze the interaction between the partner’s perceived gender (within–subjects) and the subject’s gender (between–subjects). For the (i) and (ii) analyses we will use t–tests, whereas for the (iii) analyses we will use mixed–model ANOVAs. Context 1 INTRODUCTION Pair programming is an increasingly popular collaboration paradigm that has been shown to be an effective tool in Computer Science education as measured by positive influence on grades, class performance, confidence, productivity, and motivation to stay [6], especially for women [19, 24]. In pair programming, two partners work closely together to solve a programming task. As such, their ability to engage with each other is key. However, these interactions are influenced by implicit gender bias [12, 18], such as assuming women are less technically competent [18]. This is a widely observed phenomenon even in highly–structured settings [6, 13]. Social sciences research indicates that one’s behavior of an individual is affected by the behavior of their peers [8]. Therefore, implicit gender bias based on perception of peers may have effects on one’s behavior, potentially influencing pair programming experience. In this work, in a non–colocated (i.e. remote) environment in which the gender of the peers cannot be directly observed, our goal is to explore whether Software Engineering students change their behavior when the perceived gender of their remote pair programming partners changes from man to woman or vice versa. Note that, while we recognize that many students may identify as neither men nor women, our initial exploration focuses primarily on interactions between students who identify as one of these, so that we can better align our findings with the existing literature on implicit gender bias. The potential biases in interactions involving gender-fluid, non-gender-conforming, or nonbinary students is a rich and complex topic deserving its own subsequent study. To achieve our goal, we plan to search for differences not only in the perceived productivity of pair programming compared to solo programming, the partner’s perceived technical competency compared to their own, and the partners’ perceived skill level, but also in the interaction behavior, i.e. the frequency of source code additions, deletions, validations, etc., and the type and relative frequency of dialog messages used for collaborative behavior. To get early feedback on the infrastructure supporting our proposal, we ran two pilot studies, one at each university, with a limited number of students, where we could check the comprehensibility of the questionnaires used to gather subjective data, the applicability of the message tagging (described in Section 2), and the capabilities of the twincode platform, which is briefly described below. This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv available on arxiv

Can Perceived Gender Bias Shape Pair Programming in Remote Settings?

Može li uočena rodna pristrasnost oblikovati programiranje para u udaljenim postavkama?

¿Puede el sesgo de género percibido afectar la programación por parejas en entornos remotos?

آیا تعصب جنسیتی درک شده می‌تواند برنامه‌نویسی جفتی را در تنظیمات از راه دور ایجاد کند؟

Les préjugés sexistes perçus peuvent-ils influencer la programmation en binôme dans les environnements éloignés ?

Èske yo ka pèrsuite sèks patipri fòm pè pwogramasyon nan anviwònman Remote?

認識された性別による偏見は、リモート環境でのペアプログラミングに影響を与える可能性がありますか?

შეიძლება თუ არა აღქმული გენდერული მიკერძოების ფორმის წყვილის პროგრამირება დისტანციურ პარამეტრებში?

តើអាចយល់ឃើញថា ភាពលំអៀងយេនឌ័រ ការសរសេរកម្មវិធីគូក្នុងការកំណត់ពីចម្ងាយ?

ایا په لیرې پرتو ترتیباتو کې د جنډر تعصب شکل جوړه برنامې پیژندل کیدی شي؟

O preconceito de gênero percebido pode moldar a programação em pares em ambientes remotos?

Может ли воспринимаемая гендерная предвзятость влиять на парное программирование в удаленных условиях?

Чи може сприйняте гендерне упередження формувати парне програмування у віддалених налаштуваннях?

Authors:
(1) Suzanna Sia, Johns Hopkins University;
(2) David Mueller;
(3) Kevin Duh. Table of Links Abstract and 1. Background
2. Data and Settings
3. Where does In-context MT happen?
4. Characterising Redundancy in Layers
5. Inference Efficiency
6. Further Analysis
7. Conclusion, Acknowledgments, and References
A. Appendix 2. Data and Settings Models We use GPTNEO2.7B (Black et al., 2021), BLOOM3B (Scao et al., 2022), LLAMA7B and LLAMA7Bchat (Touvron et al., 2023), the instruction-tuned variant, in all of our experiments. GPTNEO2.7B has 32 layers and 20 heads, BLOOM3B has 30 layers and 32 heads, while LLAMA7B has 32 layers and 32 heads. The checkpoints we use are from the transformers library (Wolf et al., 2019). GPTNEO was trained on The PILE (Gao et al., 2020), an 825GB text dataset which consists of roughly 98% English data. Despite being mostly monolingual, The PILE contains Europarl which GPTNEO was trained on at a document level (rather than a sentence level). Conversely, BLOOM was trained on the ROOTS corpus (Laurençon et al., 2022), a composite collection of 498 datasets that were explicitly selected to be multilingual, representing 46 natural languages and 13 programming languages. LLAMA training data consists primarily of common crawl, C4, wikipedia, stack exchange as major sources. To our knowledge, there has not been any reports of sentence level parallel corpora in the training datasets of these models. Data We test our models using FLORES (Goyal et al., 2021) en ↔ fr which we report in the main paper, and a small study on extending Section 3 to en ↔ pt in the Appendix. Prompt examples are drawn from the development set. We evaluate the generations using BLEU scores, following the implementation from Post (2018). Prompt Format Our prompts may consist of instructions, examples, both, or none. Importantly, we adopt neutral delimiters, "Q:" and "A:" to separate the prompt and the start of machine generated text. This ensures that the models do not have any information from the delimiters on what the task is. [1] When no natural language instructions are used the model input will be Q: {source_sentence} A: Instructions are given in natural language and take the form: Translate from {L1} to {L2}: Q: {source_sentence} A:, where L1 = English and L2 = French if the source and target languages are English and French respectively. Examples are given after instructions, and similarly delimited by Q: and A:. See Appendix: Table 1 for an example. This paper is available on arxiv under CC BY 4.0 DEED license. [1] In an earlier exploration, we found that supplying the model with language indicators only, e.g., "English:", "French:" was sufficient for strong models (llama7b, llama7b-chat) to perform the task without seeing any instructions or examples in the context. Authors: (1) Suzanna Sia, Johns Hopkins University; (2) David Mueller; (3) Kevin Duh. Authors: Authors: (1) Suzanna Sia, Johns Hopkins University; (2) David Mueller; (3) Kevin Duh. Table of Links Abstract and 1. Background 2. Data and Settings 3. Where does In-context MT happen? 4. Characterising Redundancy in Layers 5. Inference Efficiency 6. Further Analysis 7. Conclusion, Acknowledgments, and References A. Appendix Abstract and 1. Background Abstract and 1. Background 2. Data and Settings 2. Data and Settings 3. Where does In-context MT happen? 3. Where does In-context MT happen? 4. Characterising Redundancy in Layers 4. Characterising Redundancy in Layers 5. Inference Efficiency 5. Inference Efficiency 6. Further Analysis 6. Further Analysis 7. Conclusion, Acknowledgments, and References 7. Conclusion, Acknowledgments, and References A. Appendix A. Appendix 2. Data and Settings Models We use GPTNEO2.7B (Black et al., 2021), BLOOM3B (Scao et al., 2022), LLAMA7B and LLAMA7Bchat (Touvron et al., 2023), the instruction-tuned variant, in all of our experiments. GPTNEO2.7B has 32 layers and 20 heads, BLOOM3B has 30 layers and 32 heads, while LLAMA7B has 32 layers and 32 heads. The checkpoints we use are from the transformers library (Wolf et al., 2019). Models GPTNEO was trained on The PILE (Gao et al., 2020), an 825GB text dataset which consists of roughly 98% English data. Despite being mostly monolingual, The PILE contains Europarl which GPTNEO was trained on at a document level (rather than a sentence level). Conversely, BLOOM was trained on the ROOTS corpus (Laurençon et al., 2022), a composite collection of 498 datasets that were explicitly selected to be multilingual, representing 46 natural languages and 13 programming languages. LLAMA training data consists primarily of common crawl, C4, wikipedia, stack exchange as major sources. To our knowledge, there has not been any reports of sentence level parallel corpora in the training datasets of these models. Data We test our models using FLORES (Goyal et al., 2021) en ↔ fr which we report in the main paper, and a small study on extending Section 3 to en ↔ pt in the Appendix. Prompt examples are drawn from the development set. We evaluate the generations using BLEU scores, following the implementation from Post (2018). Prompt Format Our prompts may consist of instructions, examples, both, or none. Importantly, we adopt neutral delimiters, "Q:" and "A:" to separate the prompt and the start of machine generated text. This ensures that the models do not have any information from the delimiters on what the task is. [1] Prompt Format When no natural language instructions are used the model input will be Q: {source_sentence} A: Instructions are given in natural language and take the form: Translate from {L1} to {L2}: Q: {source_sentence} A:, where L1 = English and L2 = French if the source and target languages are English and French respectively. Examples are given after instructions, and similarly delimited by Q: and A:. See Appendix: Table 1 for an example. This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv [1] In an earlier exploration, we found that supplying the model with language indicators only, e.g., "English:", "French:" was sufficient for strong models (llama7b, llama7b-chat) to perform the task without seeing any instructions or examples in the context.

Where does In-context Translation Happen in Large Language Models: Data and Settings

Authors:
(1) Wanru Zhao, University of Cambridge, Shanghai AI Laboratory with Equal contribution;
(2) Yaxin Du, Shanghai Jiao Tong University with Equal contribution;
(3) Nicholas D. Lane, University of Cambridge and Flower Labs;
(4) Siheng Chen, Shanghai AI Laboratory and Shanghai Jiao Tong University;
(5) Yanfeng Wang, Shanghai AI Laboratory and Shanghai Jiao Tong University. Table of Links Abstract and Introduction
Motivation and Setup: How low-quality data affects the performance of Collaborative Training
Proposed Workflow for Data Quality Control
Experiments
Conclusion and Future Work, and References
A. Related Work
B. Heterogeneity Settings
C. Experimental Details
D. Ablation study of Unified Scoring with Anchor Data
E. Examples for low-and high- quality Data B HETEROGENEITY SETTINGS To model real-world scenario, we designed two heterogeneous settings: NIID-1 and NIID-2. NIID-1 replicates a typical scenario in federated learning classification tasks (Yurochkin et al., 2019; Wang et al., 2020a;b; Li et al., 2021; Shi et al., 2022), where the distribution of low-quality data among clients follows a Dirichlet distribution with parameter β = 1, while ensuring that the volume of data processed by each client remains equal. In contrast, NIID-2 addresses a skewed classification task scenario within FL (McMahan et al., 2017; Li et al., 2020), assigning 70% of low-quality data to half of the clients and 90% to the other half, yet maintaining an equal size of training data across all clients. The distributions for these settings are illustrated in Figure 3. Table2 shows the low-quality data traing and data quality control federated NIID-2 setting. This paper is available on arxiv under CC BY 4.0 DEED license. Authors: (1) Wanru Zhao, University of Cambridge, Shanghai AI Laboratory with Equal contribution; (2) Yaxin Du, Shanghai Jiao Tong University with Equal contribution; (3) Nicholas D. Lane, University of Cambridge and Flower Labs; (4) Siheng Chen, Shanghai AI Laboratory and Shanghai Jiao Tong University; (5) Yanfeng Wang, Shanghai AI Laboratory and Shanghai Jiao Tong University. Authors: Authors: (1) Wanru Zhao, University of Cambridge, Shanghai AI Laboratory with Equal contribution; (2) Yaxin Du, Shanghai Jiao Tong University with Equal contribution; (3) Nicholas D. Lane, University of Cambridge and Flower Labs; (4) Siheng Chen, Shanghai AI Laboratory and Shanghai Jiao Tong University; (5) Yanfeng Wang, Shanghai AI Laboratory and Shanghai Jiao Tong University. Table of Links Abstract and Introduction Motivation and Setup: How low-quality data affects the performance of Collaborative Training Proposed Workflow for Data Quality Control Experiments Conclusion and Future Work, and References A. Related Work B. Heterogeneity Settings C. Experimental Details D. Ablation study of Unified Scoring with Anchor Data E. Examples for low-and high- quality Data Abstract and Introduction Abstract and Introduction Motivation and Setup: How low-quality data affects the performance of Collaborative Training Motivation and Setup: How low-quality data affects the performance of Collaborative Training Proposed Workflow for Data Quality Control Proposed Workflow for Data Quality Control Experiments Experiments Conclusion and Future Work, and References Conclusion and Future Work, and References A. Related Work A. Related Work B. Heterogeneity Settings B. Heterogeneity Settings C. Experimental Details C. Experimental Details D. Ablation study of Unified Scoring with Anchor Data D. Ablation study of Unified Scoring with Anchor Data E. Examples for low-and high- quality Data E. Examples for low-and high- quality Data B HETEROGENEITY SETTINGS To model real-world scenario, we designed two heterogeneous settings: NIID-1 and NIID-2. NIID-1 replicates a typical scenario in federated learning classification tasks (Yurochkin et al., 2019; Wang et al., 2020a;b; Li et al., 2021; Shi et al., 2022), where the distribution of low-quality data among clients follows a Dirichlet distribution with parameter β = 1, while ensuring that the volume of data processed by each client remains equal. In contrast, NIID-2 addresses a skewed classification task scenario within FL (McMahan et al., 2017; Li et al., 2020), assigning 70% of low-quality data to half of the clients and 90% to the other half, yet maintaining an equal size of training data across all clients. The distributions for these settings are illustrated in Figure 3. Table2 shows the low-quality data traing and data quality control federated NIID-2 setting. This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv

Enhancing Data Quality in Federated Fine-Tuning of Foundation Models: Heterogeneity Settings

Toolify is your go-to directory for discovering the best AI tools across a wide range of categories. Whether you're looking for productivity apps, creative AI solutions, or anything in between, Toolify has it all. With a comprehensive AI tools list that's updated daily, you'll always find something new and useful. Dive into a world of innovation and enhance your digital toolkit with Toolify's expertly curated selection. The platform categorizes tools based on their features and usability, allowing users to find exactly what they need quickly. From enhancing productivity with AI-powered task managers to exploring creative design tools, Toolify offers something for everyone. Whether you are a developer, marketer, or educator, the diverse categories ensure you find the right tools to meet your specific needs. Explore more categories and tools by visiting the AI Tools Directory. Key Features of Toolify Toolify stands out as a top destination for discovering most-used AI tools and resources. With its continuous updates and user-centric approach, it caters to both tech enthusiasts and professionals seeking the latest advancements in artificial intelligence. Daily Updates and New Additions Every day, Toolify enriches its platform with new AI tools and innovations. This commitment to daily updates ensures users have access to the most current technologies, empowering them to stay ahead in an ever-evolving digital landscape. Extensive Categories for Easy Access Navigating through Toolify is seamless, thanks to its extensive range of categories. From AI chatbots to data analytics tools, each category is meticulously organized to help users find the tools they need with ease and efficiency. We also have rankings for various AI tools, such as Top AI by Categories, Regions, Sources, Revenue, and more, to help you discover the best tools tailored to your needs and goals. Explore the wide range of categories offered by Toolify to find tools tailored to your specific needs. User-Friendly Interface Toolify's interface is designed for simplicity and ease of use. Its intuitive layout allows users to explore and discover AI tools without any hassle, ensuring a smooth and enjoyable experience. Discover new tools quickly by visiting the new additions section. Target Audience for AI Tools AI tools are not just for tech enthusiasts or IT professionals. They cater to a wide range of users, each with unique needs and goals. Whether you're a student, a marketing professional, or a small business owner, AI tools can significantly enhance your daily tasks. Let's explore who can benefit the most from these tools. Students: AI education assistants can help with homework and provide personalized learning experiences. Learn more on the AI Education Assistant page.
Professionals: AI voice assistants streamline workflow by setting reminders and performing tasks quickly. Check out offerings on our AI Voice Assistants page.
E-commerce: Boost product appeal with AI product description generators that create engaging content. Discover more on our AI Product Description Generator page. These tools are designed to make life easier and productivity higher. They break down complex tasks into manageable actions, allowing more time for creativity and strategic thinking. If you're curious about how AI can enhance your work or studies, exploring these tools is a great place to start. Embrace the future of technology, where AI does the heavy lifting, and you're free to focus on what truly matters. Whether you're a developer, researcher, or just curious about AI, Toolify offers insights and tools that cater to all levels of expertise. With up-to-date information and a wide range of applications, Toolify is your go-to for advancing in AI technology. This story was distributed by Margrowth under HackerNoon’s Brand As An Author Program. Learn more about the program here: https://business.hackernoon.com/brand-as-author Toolify is your go-to directory for discovering the best AI tools across a wide range of categories. Whether you're looking for productivity apps, creative AI solutions, or anything in between, Toolify has it all. With a comprehensive AI tools list that's updated daily, you'll always find something new and useful. Dive into a world of innovation and enhance your digital toolkit with Toolify's expertly curated selection. Toolify Toolify The platform categorizes tools based on their features and usability, allowing users to find exactly what they need quickly. From enhancing productivity with AI-powered task managers to exploring creative design tools, Toolify offers something for everyone. Whether you are a developer, marketer, or educator, the diverse categories ensure you find the right tools to meet your specific needs. Explore more categories and tools by visiting the AI Tools Directory. Key Features of Toolify Key Features of Toolify Toolify stands out as a top destination for discovering most-used AI tools and resources. With its continuous updates and user-centric approach, it caters to both tech enthusiasts and professionals seeking the latest advancements in artificial intelligence. most-used AI tools most-used AI tools Daily Updates and New Additions Daily Updates and New Additions Every day, Toolify enriches its platform with new AI tools and innovations. This commitment to daily updates ensures users have access to the most current technologies, empowering them to stay ahead in an ever-evolving digital landscape. new AI tools new AI tools Extensive Categories for Easy Access Navigating through Toolify is seamless, thanks to its extensive range of categories. From AI chatbots to data analytics tools, each category is meticulously organized to help users find the tools they need with ease and efficiency. We also have rankings for various AI tools, such as Top AI by Categories , Regions, Sources, Revenue, and more, to help you discover the best tools tailored to your needs and goals. Top AI by Categories Top AI by Categories Explore the wide range of categories offered by Toolify to find tools tailored to your specific needs. User-Friendly Interface Toolify's interface is designed for simplicity and ease of use. Its intuitive layout allows users to explore and discover AI tools without any hassle, ensuring a smooth and enjoyable experience. Discover new tools quickly by visiting the new additions section. Target Audience for AI Tools AI tools are not just for tech enthusiasts or IT professionals. They cater to a wide range of users, each with unique needs and goals. Whether you're a student, a marketing professional, or a small business owner, AI tools can significantly enhance your daily tasks. Let's explore who can benefit the most from these tools. Students: AI education assistants can help with homework and provide personalized learning experiences. Learn more on the AI Education Assistant page. Professionals: AI voice assistants streamline workflow by setting reminders and performing tasks quickly. Check out offerings on our AI Voice Assistants page. E-commerce: Boost product appeal with AI product description generators that create engaging content. Discover more on our AI Product Description Generator page. Students: AI education assistants can help with homework and provide personalized learning experiences. Learn more on the AI Education Assistant page. Professionals: AI voice assistants streamline workflow by setting reminders and performing tasks quickly. Check out offerings on our AI Voice Assistants page. E-commerce: Boost product appeal with AI product description generators that create engaging content. Discover more on our AI Product Description Generator page. These tools are designed to make life easier and productivity higher. They break down complex tasks into manageable actions, allowing more time for creativity and strategic thinking. If you're curious about how AI can enhance your work or studies, exploring these tools is a great place to start. Embrace the future of technology, where AI does the heavy lifting, and you're free to focus on what truly matters. Whether you're a developer, researcher, or just curious about AI, Toolify offers insights and tools that cater to all levels of expertise. With up-to-date information and a wide range of applications, Toolify is your go-to for advancing in AI technology. This story was distributed by Margrowth under HackerNoon’s Brand As An Author Program. Learn more about the program here: https://business.hackernoon.com/brand-as-author This story was distributed by Margrowth under HackerNoon’s Brand As An Author Program. Learn more about the program here: https://business.hackernoon.com/brand-as-author This story was distributed by Margrowth under HackerNoon’s Brand As An Author Program. Learn more about the program here: https://business.hackernoon.com/brand-as-author This story was distributed by Margrowth under HackerNoon’s Brand As An Author Program. Learn more about the program here: https://business.hackernoon.com/brand-as-author https://business.hackernoon.com/brand-as-author

New Tool Can Help You Discover The Best AI Websites On the Planet

Yeni Alət Planetdəki Ən Yaxşı Süni İntellekt Vebsaytlarını Kəşf Etməkdə Sizə Kömək edə bilər

Una nueva herramienta puede ayudarte a descubrir los mejores sitios web de IA del planeta

כלי חדש יכול לעזור לך לגלות את אתרי הבינה המלאכותית הטובים ביותר על פני כדור הארץ

Novi alat može vam pomoći da otkrijete najbolja AI web mjesta na planetu

Alat Baru Dapat Membantu Anda Menemukan Situs Web AI Terbaik di Dunia

新しいツールは、地球上で最高の AI ウェブサイトを見つけるのに役立ちます

Naujas įrankis gali padėti atrasti geriausias AI svetaines planetoje

Jauns rīks var palīdzēt atklāt labākās AI vietnes uz planētas

नयाँ उपकरणले तपाईंलाई ग्रहमा सर्वश्रेष्ठ एआई वेबसाइटहरू पत्ता लगाउन मद्दत गर्न सक्छ

Новый инструмент поможет вам найти лучшие сайты с искусственным интеллектом на планете

Chishandiso Chitsva Chinogona Kukubatsira Kuti Utsvage Iyo Yakanakisa AI Webhusaiti Panyika

Authors:
(1) Xueying Mao, School of Computer Science, Fudan University, China (xymao22@m.@fudan.edu.cn);
(2) Xiaoxiao Hu, School of Computer Science, Fudan University, China (xxhu23@m.fudan.edu.cn);
(3) Wanli Peng, School of Computer Science, Fudan University, China (pengwanli@fudan.edu.cn);
(4) Zhenliang Gan, School of Computer Science, Fudan University, China (zlgan23@m.@fudan.edu.cn);
(5) Qichao Ying, School of Computer Science, Fudan University, China (qcying20@fudan.edu.cn);
(6) Zhenxing Qian, School of Computer Science, Fudan University, China and a Corresponding Author (zxqian@fudan.edu.cn);
(7) Sheng Li, School of Computer Science, Fudan University, China (lisheng@fudan.edu.cn);
(8) Xinpeng Zhang, School of Computer Science, Fudan University, China (zhangxinpeng@fudan.edu.cn). Editor's note: This is Part 6 of 7 of a study describing the development of a new method to hide secret messages in semantic features of videos, making it more secure and resistant to distortion during online sharing. Read the rest below. Table of Links Abstract and 1. Introduction
2. Related Work
3. Proposed Approach


Experiments


4.1. Experimental Setups
4.2. Performance Analysis
4.3. Ablation Study
Conclusions and References 4.3. Ablation Study Embedding Position of Secret Message. In our generation network with 9 Secret-ID blocks, we explore different positions for embedding the secret message. We divide the secret message into two 9-bit segments and allocate their positions. In detail, Setting (a): 1st-4th blocks and 5th-9th blocks. Setting (b): 1st-2nd blocks and 3rd-4th blocks. Setting (c): 5th-6th blocks and 7th-8th blocks. They are in comparison of the standard setting of RoGVS: 1st-3rd blocks and 4th-6th blocks. Table 2 displays the performance for these four setups. Both Settings b and c show a considerable decrease compared to Settings a and d, suggesting that adding more Secret-ID blocks improves performance. Notably, Setting c outperforms Setting b, indicating the higher influence of subsequent blocks on the generated image. Ablation on Attacking Layer, λ & Discriminator. Fig 5 shows even without the module, our method demonstrates considerable robustness, surpassing the three comparative methods. The addition of attacking layer improves accuracy by an average of 6%. Table 3 presents the impact of λ on the extraction accuracy. More ablation results on λ and the discriminator are displayed in the supplement. This paper is available on arxiv under CC 4.0 license. Authors: (1) Xueying Mao, School of Computer Science, Fudan University, China (xymao22@m.@fudan.edu.cn); (2) Xiaoxiao Hu, School of Computer Science, Fudan University, China (xxhu23@m.fudan.edu.cn); (3) Wanli Peng, School of Computer Science, Fudan University, China (pengwanli@fudan.edu.cn); (4) Zhenliang Gan, School of Computer Science, Fudan University, China (zlgan23@m.@fudan.edu.cn); (5) Qichao Ying, School of Computer Science, Fudan University, China (qcying20@fudan.edu.cn); (6) Zhenxing Qian, School of Computer Science, Fudan University, China and a Corresponding Author (zxqian@fudan.edu.cn); (7) Sheng Li, School of Computer Science, Fudan University, China (lisheng@fudan.edu.cn); (8) Xinpeng Zhang, School of Computer Science, Fudan University, China (zhangxinpeng@fudan.edu.cn). Authors: Authors: (1) Xueying Mao, School of Computer Science, Fudan University, China (xymao22@m.@fudan.edu.cn); (2) Xiaoxiao Hu, School of Computer Science, Fudan University, China (xxhu23@m.fudan.edu.cn); (3) Wanli Peng, School of Computer Science, Fudan University, China (pengwanli@fudan.edu.cn); (4) Zhenliang Gan, School of Computer Science, Fudan University, China (zlgan23@m.@fudan.edu.cn); (5) Qichao Ying, School of Computer Science, Fudan University, China (qcying20@fudan.edu.cn); (6) Zhenxing Qian, School of Computer Science, Fudan University, China and a Corresponding Author (zxqian@fudan.edu.cn); (7) Sheng Li, School of Computer Science, Fudan University, China (lisheng@fudan.edu.cn); (8) Xinpeng Zhang, School of Computer Science, Fudan University, China (zhangxinpeng@fudan.edu.cn). Editor's note: This is Part 6 of 7 of a study describing the development of a new method to hide secret messages in semantic features of videos, making it more secure and resistant to distortion during online sharing. Read the rest below. Editor's note: This is Part 6 of 7 of a study describing the development of a new method to hide secret messages in semantic features of videos, making it more secure and resistant to distortion during online sharing. Read the rest below. Editor's note: This is Part 6 of 7 of a study describing the development of a new method to hide secret messages in semantic features of videos, making it more secure and resistant to distortion during online sharing. Read the rest below. Editor's note: This is Part 6 of 7 of a study describing the development of a new method to hide secret messages in semantic features of videos, making it more secure and resistant to distortion during online sharing. Read the rest below. Table of Links Abstract and 1. Introduction 2. Related Work 3. Proposed Approach Experiments 4.1. Experimental Setups 4.2. Performance Analysis 4.3. Ablation Study Conclusions and References Abstract and 1. Introduction Abstract and 1. Introduction 2. Related Work 2. Related Work 3. Proposed Approach 3. Proposed Approach Experiments Experiments Experiments 4.1. Experimental Setups 4.1. Experimental Setups 4.2. Performance Analysis 4.2. Performance Analysis 4.3. Ablation Study 4.3. Ablation Study Conclusions and References Conclusions and References 4.3. Ablation Study Embedding Position of Secret Message. In our generation network with 9 Secret-ID blocks, we explore different positions for embedding the secret message. We divide the secret message into two 9-bit segments and allocate their positions. In detail, Setting (a): 1st-4th blocks and 5th-9th blocks. Embedding Position of Secret Message. Setting (b): 1st-2nd blocks and 3rd-4th blocks. Setting (c): 5th-6th blocks and 7th-8th blocks. They are in comparison of the standard setting of RoGVS: 1st-3rd blocks and 4th-6th blocks. Table 2 displays the performance for these four setups. Both Settings b and c show a considerable decrease compared to Settings a and d, suggesting that adding more Secret-ID blocks improves performance. Notably, Setting c outperforms Setting b, indicating the higher influence of subsequent blocks on the generated image. Ablation on Attacking Layer, λ & Discriminator. Fig 5 shows even without the module, our method demonstrates considerable robustness, surpassing the three comparative methods. The addition of attacking layer improves accuracy by an average of 6%. Table 3 presents the impact of λ on the extraction accuracy. More ablation results on λ and the discriminator are displayed in the supplement. Ablation on Attacking Layer, λ & Discriminator. This paper is available on arxiv under CC 4.0 license. This paper is available on arxiv under CC 4.0 license. available on arxiv

Scientists Find Placement and Precision Matter in Hiding Secret Messages in Videos

Here's How Developers Can Ensure Sensitive Data Does Not Accidentally Reach Another Device

Así pueden los desarrolladores garantizar que los datos confidenciales no lleguen accidentalmente a otro dispositivo

Így biztosíthatják a fejlesztők, hogy az érzékeny adatok véletlenül ne érjenek el egy másik eszközt

Ահա թե ինչպես են ծրագրավորողները կարող համոզվել, որ զգայուն տվյալները պատահաբար չեն հասնում մեկ այլ սարքի

開発者が機密データが誤って別のデバイスに届かないようにする方法

აი, როგორ შეუძლიათ დეველოპერებს დარწმუნდნენ, რომ მგრძნობიარე მონაცემები შემთხვევით არ მოხვდება სხვა მოწყობილობაზე

នេះជារបៀបដែលអ្នកអភិវឌ្ឍន៍អាចធានាថាទិន្នន័យរសើបមិនទៅដល់ឧបករណ៍មួយផ្សេងទៀតដោយចៃដន្យ

Mona ke Kamoo Bahlami ba ka Netefatšago gore Datha ye e nago le Bohloko ga e Fihlelele Sedirišwa se Seng ka Phošo

Iată cum dezvoltatorii se pot asigura că datele sensibile nu ajung accidental la alt dispozitiv

Heano Mashandisiro Anogona Kuita Vagadziri Vese Sensitive Dhata Haisviki Netsaona Imwe Chishandiso

Waa kan sida ay horumariyayaashu u hubin karaan in xogta xasaasiga ahi aanay si lama filaan ah u gaadhin qalab kale

Ось як розробники можуть гарантувати, що конфіденційні дані випадково не потраплять на інший пристрій

Nantsi indlela abaPhuhlisi abanokuqinisekisa ngayo ukuba iDatha enovakalelo ayifikeleli kwesinye isixhobo ngengozi

Stories, News & Mentions about SEATTI

Seatti’s Stories on HackerNoon

Seatti's News & Mentions