Hostile Neural Networks in FTB Skies 2: Aero

FTB Skies 2: Aero 1.13.0 ships version 6.5.1 · First seen in 1.0.3

Checked Sep 17, 2026FTB Skies 2: Aero 1.13.0Sources: FTB Skies 2: Aero 1.13.0 modlist.json · Modrinth

Report a mistake on this page

Also in 8 other packs →

Hostile Neural Networks is a mod in FTB Skies 2: Aero 1.13.0, where it is version 6.5.1. Simulate Mobs to create not-so-simulated drops! (Summary from the mod's Modrinth page.)

Hostile Neural Networks guide for FTB Skies 2: Aero

Written for FTB Skies 2: Aero 1.10.0 · also fits 1.9.0 to 1.9.1 · updated Oct 14, 2026

Last verified for FTB Skies 2: Aero v1.10.0 on Oct 14, 2026 · text last changed Oct 14, 2026

This is the newest guide. It hasn't been checked against 1.13.0 yet.

Also applies to:FTB Evolution 2 1.0.2 to 1.4.0

Sections
8
Recipes shown
12
Items named
12
Quests covered
4

Every recipe and number here was checked against FTB Skies 2: Aero 1.10.0, read from the running game on Sep 22, 2026.

What is Hostile Neural Networks?

1 part

Hostile Neural Networks lets you simulate mobs and turn the simulations into their drops. You build a Data Model of a mob, train it, run simulations, and feed the resulting predictions into a machine that spits out loot. The pack includes version 6.5.1 of the mod.[1][2]

Here is the flow. You craft a Model Framework and use it on the target mob to build its Data Model. Models start out Faulty and must be trained up to Basic in a Deep Learner. Once a model is Basic or higher, you run it in the Simulation Chamber, which needs energy and a catalyst, usually a Prediction Matrix. Accuracy rises with the model tier, up to 99.5% at Self Aware. Finally, the Loot Fabricator uses power to convert mob predictions into drops.[1]

Where it sits in the quest book

The pack's quest book covers this mod in the "Generating Resources" chapter. That chapter is in the "Resources & Logistics" group and is chapter 8 of 31 in the book. It has 99 quests and sits alongside other resource farming topics such as Mystical Agriculture, Productive Bees and Spawner Farms.[3]

The mod's quests form one chain inside that chapter:[3]

  1. "Hostile Neural Networks!" quest is the starting quest. It asks for the Model Framework and the Deep Learner, and tells you to level the model to at least Basic.[3]
  2. a quest in "Generating Resources" comes after it. It covers the Simulation Chamber and Prediction Matrices, including its Training Mode and Inference Mode. The book has no title of its own for this quest, so it is named after the item it asks for.[3]
  3. a quest in "Generating Resources" comes after that. It explains choosing a drop in the Loot Fabricator and its Queue Mode. This quest is also untitled in the book.[3]
  4. "Neural Data Center" quest comes after the Loot Fabricator quest. It describes the Data Center multiblock, a 7x7x7 structure that runs up to 25 Self Aware data models at the same time, with Data Center IO Port blocks for energy, inputs and outputs.[3]

The text of 82 of the chapter's 99 quests is in the quest book itself, so check there for anything beyond this chain.[3]

Unverified Whether the ATM Star path uses this mod is not shown by the sources here. Check the pack's quest book for that.

How do you get started with Hostile Neural Networks?

5 parts

Hostile Neural Networks (HNN for short) turns mob kills into loot through Data Models, simulations and Predictions. In this pack's quest book it lives in the "Generating Resources" chapter (group "Resources & Logistics"), and it starts with the quest "Hostile Neural Networks!" quest. Here is the order I'd make things in, and why.[1][3]

Step one: the Model Framework

Start with the Model Framework. Right-click it on a mob and it builds that mob's Data Model. Attuning this way is switched on in the pack's config ([[config:config/hostilenetworks.cfg#models.Right Click To Attune]]), the same as the mod default. The pack changed this recipe, so use the one below, not one from another guide.[3][4][5]

Model FrameworkChanged by the pack

Crafting Shaped

Crafting Shaped
Clay BallRepeaterClay Ball
RedstoneSmooth StoneRedstone
Clay BallGold IngotClay Ball

Recipe id: hostilenetworks:framework

This recipe is changed by the pack.

Step two: the Deep Learner

Next, make the Deep Learner and put your Data Model inside. Every kill of that mob improves the model. All models start out Faulty, and you need to get yours to at least Basic before it can be used for Predictions. The pack changed this recipe too.[1][3][4]

Deep LearnerChanged by the pack

Crafting Shaped

Crafting Shaped
any obsidianRepeaterany obsidian
Repeaterany glass panesRepeater
any obsidianRedstoneany obsidian

Recipe id: hostilenetworks:deep_learner

This recipe is changed by the pack.

Step three: Simulation Chamber and Prediction Matrix

Once a model is Basic or higher, you can stop hunting by hand. Place it in a Simulation Chamber and feed it power and Prediction Matrix items. A failed simulation gives you a Generalized Prediction, and a success gives that mob's own Prediction. Both recipes are changed by the pack, and one craft of the matrix makes 16.[1][3][4]

Simulation ChamberChanged by the pack

Crafting Shaped

Crafting Shaped
Empty slotany glass panesEmpty slot
Ender Pearlany obsidianEnder Pearl
Lapis LazuliComparatorLapis Lazuli

Recipe id: hostilenetworks:sim_chamber

This recipe is changed by the pack.

16× Prediction MatrixChanged by the pack

Crafting Shaped

Crafting Shaped
Iron Ingotany glass panesEmpty slot
any glass panesClay Ballany glass panes
Empty slotany glass panesGold Ingot

Recipe id: hostilenetworks:prediction_matrix

This recipe is changed by the pack and makes 16 matrices per craft.

The chamber has two modes. Training Mode runs simulations that upgrade the model's data, and Inference Mode produces Predictions but doesn't upgrade the model. The quest a quest in "Generating Resources" covers this and comes right after the first HNN quest.[3][6]

Step four: the Loot Fabricator

Predictions only become real items in the Loot Fabricator. If a mob can drop several things, you pick which one to farm in its interface. If you automate it and no choice has been made, the machine stops working. In the pack it costs 256 FE/t to run ([[config:config/hostilenetworks.cfg#power.Loot Fab Power Cost]]), the same as the mod default. The quest a quest in "Generating Resources" comes after the Generalized Ender Prediction quest, and the pack changed this recipe.[3][4][5]

Loot FabricatorChanged by the pack

Crafting Shaped

Crafting Shaped
Empty slotNetherite IngotEmpty slot
Diamondany obsidianDiamond
Gold IngotComparatorGold Ingot

Recipe id: hostilenetworks:loot_fabricator

This recipe is changed by the pack.

Later: the Data Center

When you have several models and want to scale up, the Data Center runs up to 25 Self Aware models at once. It needs a 7x7x7 hollow shell with an obsidian floor and black stained glass walls and ceiling, plus Data Center IO Port blocks set to different modes. The quest "Neural Data Center" quest walks you through building it and comes after the Loot Fabricator quest.[3][4][6]

Data CenterMod recipe

Crafting Shaped

Crafting Shaped
Echo ShardNether StarEcho Shard
Simulation Chamberany obsidianLoot Fabricator
any obsidianany obsidianany obsidian

Recipe id: hostilenetworks:data_center

This recipe is changed by the pack.

Tip: The pack raised the Data Center's power cap to 20,000,000 FE (the mod default is 8,000,000) and lowered its sim cost multiplier to 1.25 (the mod default is 1.5), so it is a bit cheaper to run here than in the base mod.[5]

How does Hostile Neural Networks work?

6 parts

Hostile Neural Networks turns mobs into loot through simulation. You make a Data Model from a Model Framework by using it on a target, train that model, run simulations with it, then turn the resulting predictions into real drops. Here is how each step works in this pack.[1]

Making and training Data Models

Every model starts in the Faulty state and has to be trained up to Basic in the Deep Learner. The Deep Learner holds a few models, and the kills you make train them. If you want its HUD somewhere else on screen, [[config:config/hostilenetworks.cfg#client.Deep Learner HUD Anchor Point]], [[config:config/hostilenetworks.cfg#client.Deep Learner HUD X Offset]] and [[config:config/hostilenetworks.cfg#client.Deep Learner HUD Y Offset]] control it. The pack leaves these at the mod defaults (top_left, 0 and 0).[1][5]

The pack keeps the main training rules at their defaults. [[config:config/hostilenetworks.cfg#models.Right Click To Attune]] is true, so right clicking a blank model on a mob or block attunes it. [[config:config/hostilenetworks.cfg#models.Action Upgrades Model]] is true, so killing mobs and breaking blocks upgrades the matching model. [[config:config/hostilenetworks.cfg#models.Continuous Accuracy]] is true, so a model's accuracy rises as it progresses toward the next tier instead of jumping only at tier changes.[5]

One setting differs from the mod default. [[config:config/hostilenetworks.cfg#models.Enable Block Data Models]] is true in this pack (the mod default is false), so block data models such as the built-in ore models are loaded. The config describes this as an experimental feature.[5]

Model FrameworkChanged by the pack

Crafting Shaped

Crafting Shaped
Clay BallRepeaterClay Ball
RedstoneSmooth StoneRedstone
Clay BallGold IngotClay Ball

Recipe id: hostilenetworks:framework

The pack changed this recipe.

Deep LearnerChanged by the pack

Crafting Shaped

Crafting Shaped
any obsidianRepeaterany obsidian
Repeaterany glass panesRepeater
any obsidianRedstoneany obsidian

Recipe id: hostilenetworks:deep_learner

The pack changed this recipe too.

The Simulation Chamber

Once a model is Basic or higher, the Simulation Chamber can run simulations with it. Every simulation produces a generalized prediction, and it may also produce a Mob Prediction depending on the model's accuracy. Accuracy rises with the tier, up to 99.5% at Self Aware. A simulation needs energy and a catalyst item, which is usually a Prediction Matrix.[1]

The chamber stores up to 2,000,000 FE, set by [[config:config/hostilenetworks.cfg#power.Sim Chamber Power Cap]], which is the mod default.[5]

It has two modes. In training mode, simulations increase the model's data but produce no outputs. In inference mode, simulations produce output but the models gain no data.[6]

How much training mode upgrades a model is set by [[config:config/hostilenetworks.cfg#models.Sim Chamber Upgrades Model]]. The pack uses 1, which means yes. A value of 0 disables it and also turns training mode off entirely, and 2 only upgrades up to tier boundaries.[5]

Simulation ChamberChanged by the pack

Crafting Shaped

Crafting Shaped
Empty slotany glass panesEmpty slot
Ender Pearlany obsidianEnder Pearl
Lapis LazuliComparatorLapis Lazuli

Recipe id: hostilenetworks:sim_chamber

The pack changed this recipe.

16× Prediction MatrixChanged by the pack

Crafting Shaped

Crafting Shaped
Iron Ingotany glass panesEmpty slot
any glass panesClay Ballany glass panes
Empty slotany glass panesGold Ingot

Recipe id: hostilenetworks:prediction_matrix

The pack changed this recipe.

Predictions and the Loot Fabricator

The Loot Fabricator uses power to convert predictions into mob drops. The available drops come from a predefined list, and you pick a target for each type of model. The fabricator remembers your choice for each type, so one fabricator can take several kinds of predictions.[1]

In this pack it costs 256 FE/t and stores up to 1,000,000 FE. Those are [[config:config/hostilenetworks.cfg#power.Loot Fab Power Cost]] and [[config:config/hostilenetworks.cfg#power.Loot Fab Power Cap]], both at the mod defaults.[5]

A Fabrication Directive stores the selections from a Loot Fabricator. Right-click to copy a layout, and Shift-Right-Click to apply it to another fabricator.[6]

You can also move a generalized prediction up a dimension by crafting. A Generalized Overworld Prediction becomes a Generalized Nether Prediction, and a Nether one becomes an Generalized Ender Prediction.[4]

Loot FabricatorChanged by the pack

Crafting Shaped

Crafting Shaped
Empty slotNetherite IngotEmpty slot
Diamondany obsidianDiamond
Gold IngotComparatorGold Ingot

Recipe id: hostilenetworks:loot_fabricator

The pack changed this recipe.

Crafting Shaped

Crafting Shaped
Empty slotany generalized predictionEmpty slot
PaperPrediction MatrixPaper
Empty slotPaperEmpty slot

Recipe id: hostilenetworks:fab_directive

Crafting Shaped

Crafting Shaped
Empty slotGeneralized Overworld PredictionEmpty slot
Generalized Overworld PredictionNetherrackGeneralized Overworld Prediction
Empty slotGeneralized Overworld PredictionEmpty slot

Recipe id: hostilenetworks:living_matter/overworld_to_nether

Crafting Shaped

Crafting Shaped
Empty slotGeneralized Nether PredictionEmpty slot
Generalized Nether PredictionEnd StoneGeneralized Nether Prediction
Empty slotGeneralized Nether PredictionEmpty slot

Recipe id: hostilenetworks:living_matter/nether_to_ender

The Data Center multiblock

The Data Center is the top-end option. It runs up to 25 Self Aware Data Models at once, and each model produces one drop per cycle into a shared output buffer.[6]

You build a hollow 7×7×7 shell around it. The floor must be obsidian, and the walls and ceiling must be black stained glass. The controller replaces one bottom-wall block and must face outwards.[6]

Insert Data Models into the 5×5 grid and matching items into the four input slots. Each active slot consumes one input per cycle and produces the drop chosen by its Fabrication Directive.[6]

IO Ports

A Data Center IO Port can replace any wall or ceiling block in the shell, and the controller detects it automatically. Right-click it to cycle through four modes: Energy accepts power, while Models, Inputs and Outputs expose the matching item slots.[6]

Data Center power in this pack

The pack raised [[config:config/hostilenetworks.cfg#power.Data Center Power Cap]] to 20,000,000 FE (the mod default is 8,000,000). The config says this is sized for 25 concurrent self-aware simulations.[5]

The pack also lowered [[config:config/hostilenetworks.cfg#models.Data Center Sim Cost Multiplier]] to 1.25 (the mod default is 1.5). That is the per-tick power cost the Data Center pays for each active model, as a multiplier on the model's base simulation cost, so this pack's Data Center is cheaper to run than the default.[5]

Data CenterMod recipe

Crafting Shaped

Crafting Shaped
Echo ShardNether StarEcho Shard
Simulation Chamberany obsidianLoot Fabricator
any obsidianany obsidianany obsidian

Recipe id: hostilenetworks:data_center

The pack changed this recipe.

How do you progress through Hostile Neural Networks?

4 parts

Hostile Neural Networks has a short, clear progression path in this pack, and the quest book walks you through it. It sits in the Generating Resources chapter (chapter 8 of 31, in the Resources & Logistics group). The four quests that matter run in a line: "Hostile Neural Networks!" quest, then a quest in "Generating Resources", then a quest in "Generating Resources", then "Neural Data Center" quest.[3]

Step 1: Data models and the Deep Learner

The first quest has you make a Model Framework and right-click it on a mob to gather that mob's Data Model. Then you craft a Deep Learner and put the model inside. From then on, every kill of that mob improves the model. To make Predictions with it, the model has to reach at least the Basic tier.[3]

Model FrameworkChanged by the pack

Crafting Shaped

Crafting Shaped
Clay BallRepeaterClay Ball
RedstoneSmooth StoneRedstone
Clay BallGold IngotClay Ball

Recipe id: hostilenetworks:framework

Deep LearnerChanged by the pack

Crafting Shaped

Crafting Shaped
any obsidianRepeaterany obsidian
Repeaterany glass panesRepeater
any obsidianRedstoneany obsidian

Recipe id: hostilenetworks:deep_learner

Step 2: Simulation Chamber and Predictions

The second quest asks for a Simulation Chamber, a Prediction Matrix and one of the Generalized Predictions. Put a Basic-tier (or better) Data Model in the chamber, feed it Prediction Matrices and power, and it runs simulations as if you were killing that mob. The power cost per simulation depends on the mob's rarity.[3]

A failed simulation gives you a Generalized Prediction, and a successful one gives you that mob's own Prediction. The chamber has two modes: Training Mode upgrades the model's data, while Inference Mode produces predictions without upgrading the model.[3]

Simulation ChamberChanged by the pack

Crafting Shaped

Crafting Shaped
Empty slotany glass panesEmpty slot
Ender Pearlany obsidianEnder Pearl
Lapis LazuliComparatorLapis Lazuli

Recipe id: hostilenetworks:sim_chamber

16× Prediction MatrixChanged by the pack

Crafting Shaped

Crafting Shaped
Iron Ingotany glass panesEmpty slot
any glass panesClay Ballany glass panes
Empty slotany glass panesGold Ingot

Recipe id: hostilenetworks:prediction_matrix

This recipe makes 16 Prediction Matrices at a time.

The Generalized Predictions come in three kinds: Generalized Overworld Prediction, Generalized Nether Prediction and Generalized Ender Prediction. You can step an Overworld one up to Nether, and a Nether one up to Ender, with these two recipes.[3][4]

Crafting Shaped

Crafting Shaped
Empty slotGeneralized Overworld PredictionEmpty slot
Generalized Overworld PredictionNetherrackGeneralized Overworld Prediction
Empty slotGeneralized Overworld PredictionEmpty slot

Recipe id: hostilenetworks:living_matter/overworld_to_nether

Crafting Shaped

Crafting Shaped
Empty slotGeneralized Nether PredictionEmpty slot
Generalized Nether PredictionEnd StoneGeneralized Nether Prediction
Empty slotGeneralized Nether PredictionEmpty slot

Recipe id: hostilenetworks:living_matter/nether_to_ender

Step 3: Loot Fabricator

The Loot Fabricator is the machine that turns a mob's Predictions into actual loot when powered. Some mobs drop more than one loot type, so you pick which one to farm in its interface. It remembers that choice when you automate it, and if no choice has been made it stops working. You can also switch it to Queue Mode to cycle through a list of outputs instead of one fixed drop.[3]

Loot FabricatorChanged by the pack

Crafting Shaped

Crafting Shaped
Empty slotNetherite IngotEmpty slot
Diamondany obsidianDiamond
Gold IngotComparatorGold Ingot

Recipe id: hostilenetworks:loot_fabricator

Step 4: The Neural Data Center

The last quest in the line is the Data Center multiblock. It is a 7x7x7 structure that runs up to 25 Self Aware data models at the same time. You need the Data Center block and Data Center IO Port blocks, plus obsidian and black stained glass for the shell.[3]

  1. Build a 7x7 obsidian platform at the bottom.[3]
  2. Raise black stained glass walls and leave the inside hollow.[3]
  3. On the outer bottom layer, place the Data Center block and 3 IO Ports, each set to a different mode: Energy, Inputs and Outputs.[3]
  4. Optionally add a fourth Port set to Models for automatic input of data models. It is not required.[3]

If the multiblock is built correctly, a blue outline shows around it. Otherwise, click the Data Center block for more details.[3]

Data CenterMod recipe

Crafting Shaped

Crafting Shaped
Echo ShardNether StarEcho Shard
Simulation Chamberany obsidianLoot Fabricator
any obsidianany obsidianany obsidian

Recipe id: hostilenetworks:data_center

Watch out: The pack changes several recipes in this chain: the Model Framework, Deep Learner, Prediction Matrix, Simulation Chamber, Loot Fabricator and Data Center IO Port. Follow the recipes shown here, not ones you remember from the mod's defaults. The Data Center and the prediction conversion recipes are the mod's own.[4]

How does Hostile Neural Networks work with other mods?

2 parts

Hostile Neural Networks connects to the rest of the pack in two ways: a pair of recipes the pack adds, and the place it has in the quest book. The mod's own project page doesn't list any other mods it supports, so everything below comes from the pack itself.[1][3][4]

Mob Tears from Predictions

The pack adds two recipes that turn the mod's Generalized Predictions (Generalized Overworld Prediction, Generalized Nether Prediction and Generalized Ender Prediction) into Mob Tears. These recipes aren't part of the mod itself.[4]

Simulations always produce a Generalized Prediction, so you get a steady supply of them whenever you run your Simulation Chamber. That makes these recipes a handy use for them.[1]

The first recipe is a Create mixing recipe. It makes 1,000 Mob Tears per batch.[4]

1000× Mob TearsAdded by the pack

Mixing (create)

Recipe id: ftb:create/mixing/mob_tears

The second is a Modern Industrialization mixer recipe. It also asks for 1,000 of an item or fluid from the experience tag, so pick whichever route suits your base.[4]

Mob TearsAdded by the pack

Mixer (modern_industrialization)

Recipe id: ftb:modern_industrialization/mixer/mob_tears

Where it sits in the quest book

The mod's quests live in the "Generating Resources" chapter. It is chapter 8 of 31, in the "Resources & Logistics" group. The chapter introduces itself as a guide to the mods that help you farm millions of items, so you'll see Hostile Neural Networks next to other resource-generation mods, including Mob Grinding Utils and Mystical Agriculture.[3]

The mod's quests form a short chain. It starts with "Hostile Neural Networks!" quest, which introduces the Model Framework and the Deep Learner. Then comes a quest in "Generating Resources", which covers the Simulation Chamber and the Prediction Matrix. After that is a quest in "Generating Resources", and the chain ends with "Neural Data Center" quest.[3]

The Neural Data Center quest is the only one in the chain whose tasks include non-mod items. It asks for the Data Center, the Data Center IO Port, Obsidian and Black Stained Glass, because the multiblock is built from a 7x7 Obsidian platform with Black Stained Glass walls.[3]

Tip: The chapter has 99 quests, and the book text for most of them isn't covered here. If you want to know how the mod fits with the other farming mods in the chapter, browse the rest of "Generating Resources" in your quest book.[3]

Which FTB Skies 2: Aero quests use Hostile Neural Networks?

2 parts

All of the quests for this mod sit in one chapter of the pack's quest book: Generating Resources. It is chapter 8 of 31, in the group Resources & Logistics, and it holds 99 quests that cover many other resource mods too. Hostile Neural Networks gets a short chain of four quests, and each one comes after the previous one.[3]

The quest chain, in order

  1. "Hostile Neural Networks!" quest is the start of the chain. It asks for a Model Framework and a Deep Learner. You right-click the Model Framework on a mob to gather its Data Model, then put that model in the Deep Learner. Every kill of that mob then improves the model. To make Predictions from it, you need to level it up to at least the Basic tier.[3]
  2. a quest in "Generating Resources" comes after the first quest. It asks for a Simulation Chamber, a Prediction Matrix, and one of the Overworld, Nether or End predictions (Generalized Overworld Prediction, Generalized Nether Prediction or Generalized Ender Prediction).[3]
  3. a quest in "Generating Resources" comes after the Prediction quest. It asks for the Loot Fabricator.[3]
  4. "Neural Data Center" quest comes after the Loot Fabricator quest. It asks for the Data Center, the Data Center IO Port, Obsidian and Black Stained Glass.[3]

What each quest teaches

The Simulation Chamber needs a Data Model that is at least Basic tier. Give it Prediction Matrices and power and it runs simulations as if you were killing that mob. The power per simulation depends on the rarity of the mob. A failed simulation gives you a Generalized Prediction, and a success gives you that mob's own Prediction. It has two modes: Training Mode upgrades the model's data, while Inference Mode produces predictions without upgrading the model.[3]

The Loot Fabricator turns Mob Predictions into actual loot when powered. If a mob can drop more than one type of loot, you pick which one to farm in its interface. When you automate it, it remembers your choice, but if no choice has been made the machine stops working. You can also switch it to Queue Mode to cycle through a list of outputs instead of producing one fixed drop.[3]

The Data Center is a 7x7x7 multiblock that runs up to 25 Self Aware data models at the same time. You build it with a 7x7 Obsidian platform at the bottom and Black Stained Glass walls, leaving the inside hollow. The Data Center block and 3 IO Ports go on the outer bottom layer, each set to a different mode: Energy, Inputs and Outputs. You can optionally add a fourth port set to Models for automatic input of data models, but it is not required.[3]

Tip: If you build the Data Center correctly, a blue outline shows around the multiblock. If it does not, click the main block of the structure for more details. The quest also points to the Data Center GuideMe for an example structure.[3]

Which Hostile Neural Networks config options matter?

4 parts

Everything here lives in config/hostilenetworks.cfg. This pack changes only a few values, and all of them are about the Data Center and the models it can load. The rest sits at the mod defaults, so you don't need to go digging unless you want to tweak things.[5]

What the pack changed

  • [[config:config/hostilenetworks.cfg#power.Data Center Power Cap]] is 20,000,000 in this pack, up from the mod default of 8,000,000. It is the maximum FE the Data Center can store, and the config note says it is sized for 25 concurrent self-aware simulations.[5]
  • [[config:config/hostilenetworks.cfg#models.Data Center Sim Cost Multiplier]] is 1.25 in this pack, down from the mod default of 1.5. It multiplies the base simCost of each active model to set the per-tick power the Data Center pays, so running models costs a little less here than in the default setup. The allowed range is 0.01 to 100.0.[5]
  • [[config:config/hostilenetworks.cfg#models.Enable Block Data Models]] is true in this pack, while the mod default is false. This loads block data models, such as the built-in ore models. The config calls it an experimental feature, and the value is not synced because it governs which models load during datapack reading.[5]

Power settings left at default

  • [[config:config/hostilenetworks.cfg#power.Sim Chamber Power Cap]] is 2,000,000 FE, the maximum stored in the Simulation Chamber.[5]
  • [[config:config/hostilenetworks.cfg#power.Loot Fab Power Cap]] is 1,000,000 FE, the maximum stored in the Loot Fabricator.[5]
  • [[config:config/hostilenetworks.cfg#power.Loot Fab Power Cost]] is 256 FE/t, the running cost of the Loot Fabricator.[5]

Model training settings left at default

  • [[config:config/hostilenetworks.cfg#models.Action Upgrades Model]] is true, so killing mobs and breaking blocks upgrades the data on the matching model.[5]
  • [[config:config/hostilenetworks.cfg#models.Right Click To Attune]] is true, so right clicking a blank data model on a mob or block attunes it to that target.[5]
  • [[config:config/hostilenetworks.cfg#models.Continuous Accuracy]] is true, so a model's accuracy rises as it gains progress toward the next tier instead of staying at the base accuracy of its current tier.[5]
  • [[config:config/hostilenetworks.cfg#models.Sim Chamber Upgrades Model]] is 1, meaning the Simulation Chamber's Training Mode upgrades model data. The options are 0 (disabled, which also turns off Training Mode entirely), 1 (yes) and 2 (only up to tier boundaries).[5]
Watch out: If you ever turn off Action Upgrades Model or Right Click To Attune, the config itself warns you to give players another way to get non-Faulty or attuned models. Otherwise you can get stuck.[5]

Deep Learner HUD

The Deep Learner HUD is a client-side setting. It uses [[config:config/hostilenetworks.cfg#client.Deep Learner HUD Anchor Point]] set to "top_left", with [[config:config/hostilenetworks.cfg#client.Deep Learner HUD X Offset]] and [[config:config/hostilenetworks.cfg#client.Deep Learner HUD Y Offset]] both at 0, which are the mod defaults. Both offsets accept values from -1000 to 1000, so you can nudge the HUD wherever you like.[5]

What changed in Hostile Neural Networks between versions?

FTB Skies 2: Aero 1.10.0 ships Hostile Neural Networks 6.5.1 (the HostileNeuralNetworks-1.21.1-6.5.1.jar file). The mod has been in the pack since Aero 1.0.3 or earlier, so it has been part of the pack for its whole run.[2]

The mod's own changelog has entries for 6.5.1, 6.5.0 and 6.4.2.[7][8][9]

Unverified The contents of those changelog entries were not available to me, so I can't tell you what changed between 6.4.2, 6.5.0 and 6.5.1. The pack changelogs also have no entries for this mod, so there is no confirmed record of pack-specific changes to it. Check the mod's own changelog before relying on any version-specific behavior.

Sources

  1. Hostile Neural Networks project page (modrinth) · official mod data
  2. FTB Skies 2: Aero 1.10.0 mod list · pack file
  3. FTB Skies 2: Aero 1.10.0 quests: "Generating Resources" · pack file
  4. FTB Skies 2: Aero 1.10.0 recipe dump: recipes making Hostile Neural Networks items · pack recipe dump
  5. FTB Skies 2: Aero 1.10.0 config: config/hostilenetworks.cfg · pack file
  6. hostilenetworks names, screen text and descriptions (English lang file) · official mod data
  7. Hostile Neural Networks 6.5.1 changelog · changelog
  8. Hostile Neural Networks 6.5.0 changelog · changelog
  9. Hostile Neural Networks 6.4.2 changelog · changelog
Where this guide comes from

Written from what FTB Skies 2: Aero itself ships: its files and recipes, the in-game guidebooks and the mods' own pages. Every statement links to the source it came from, and the numbers are checked against the pack before publishing.

Guide last verified Oct 14, 2026.

Guides for other versions: 1.0.3 to 1.8.0

Was this helpful?

Report wrong info

A few words are enough, up to 500 characters. Please do not include personal details.

Version in each release

21 releases, first in 1.0.3
FTB Skies 2: AeroVersionJarWhere it came from
1.13.0
Sep 28, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.12.1
Sep 22, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.12.0
Sep 21, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.10.0
Sep 8, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.9.1
Sep 2, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.9.0
Sep 1, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.8.0
Aug 25, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.7.3
Aug 19, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.7.2
Aug 18, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.7.1
Aug 18, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.6.1
Aug 13, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.6.0
Aug 11, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.5.1
Aug 8, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.4.1
Aug 4, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.4.0
Aug 3, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.3.0
Jul 30, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.2.1
Jul 28, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.2.0
Jul 28, 2026
6.5.1HostileNeuralNetworks-1.21.1-6.5.1.jarFrom the published release
1.1.1
Jul 25, 2026
6.5.0HostileNeuralNetworks-1.21.1-6.5.0.jarFrom the published release
1.1.0
Jul 24, 2026
6.5.0HostileNeuralNetworks-1.21.1-6.5.0.jarFrom the published release
1.0.3
Jul 17, 2026
6.4.2HostileNeuralNetworks-1.21.1-6.4.2.jarFrom the published release

"From pack files" means the list came from the pack's own files; "from the published release" means ForgeBase read the jars of the release itself.

Items and blocks

14 items from the 1.10.0 recipe dump.

FTB Skies 2: Aero quests (4)

Quests from the pack's FTB Quests book that ask for this mod's items.

See all 4 quests →

Generating Resources (4)

Config options (how FTB Skies 2: Aero sets them)

13 settings shipped in the pack's config files; 3 differ from the mod's documented default.

See all 13 settings →

Show settings
SettingFTB Skies 2: Aero valueMod default
models.Data Center Sim Cost Multiplier changed by FTB Skies 2: Aero
Per-tick power cost the Data Center pays per active model, as a multiplier on the model's base simCost. 1.5 = 150% of normal.
config/hostilenetworks.cfg
1.251.5
models.Enable Block Data Models changed by FTB Skies 2: Aero
If true, block data models (such as the built-in ore models) are loaded. This is an experimental feature. This value is not synced; it governs which models load during datapack reading.
config/hostilenetworks.cfg
truefalse
power.Data Center Power Cap changed by FTB Skies 2: Aero
The maximum FE stored in the Data Center. Sized for 25 concurrent self-aware simulations.
config/hostilenetworks.cfg
200000008000000
client.Deep Learner HUD Anchor Point
The anchor point for this element.
config/hostilenetworks.cfg
top_lefttop_left
client.Deep Learner HUD X Offset
The X offset for this element.
config/hostilenetworks.cfg
00
client.Deep Learner HUD Y Offset
The Y Offset for this element.
config/hostilenetworks.cfg
00

Where does Hostile Neural Networks appear in the FTB Skies 2: Aero quest book?

Comments and tips

Loading comments...