remove support for extras (flash attention, iQ quants

Editor/LLMEditor.cs

@@ -111,7 +111,6 @@ public override void AddModelSettings(SerializedObject llmScriptSO)
             if (llmScriptSO.FindProperty("advancedOptions").boolValue)
             {
                 attributeClasses.Add(typeof(ModelAdvancedAttribute));
-                if (LLMUnitySetup.FullLlamaLib) attributeClasses.Add(typeof(ModelExtrasAttribute));
             }
             ShowPropertiesOfClass("", llmScriptSO, attributeClasses, false);
             Space();
@@ -445,18 +444,12 @@ private void CopyToClipboard(string text)
             te.Copy();
         }
 
-        public void AddExtrasToggle()
-        {
-            if (ToggleButton("Use extras", LLMUnitySetup.FullLlamaLib)) LLMUnitySetup.SetFullLlamaLib(!LLMUnitySetup.FullLlamaLib);
-        }
-
         public override void AddOptionsToggles(SerializedObject llmScriptSO)
         {
             AddDebugModeToggle();
 
             EditorGUILayout.BeginHorizontal();
             AddAdvancedOptionsToggle(llmScriptSO);
-            AddExtrasToggle();
             EditorGUILayout.EndHorizontal();
             Space();
         }

README.md

@@ -499,8 +499,7 @@ Save the scene, run and enjoy!
 ### LLM Settings
 
 - `Show/Hide Advanced Options` Toggle to show/hide advanced options from below
-- `Log Level` select how verbose the log messages are
-- `Use extras` select to install and allow the use of extra features (flash attention and IQ quants)
+- `Log Level` select how verbose the log messages arequants)
 
 #### 💻 Setup Settings
 
@@ -551,7 +550,6 @@ If the user's GPU is not supported, the LLM will fall back to the CPU
   - `Chat Template` the chat template being used for the LLM
   - `Lora` the path of the LoRAs being used (relative to the Assets/StreamingAssets folder)
   - `Lora Weights` the weights of the LoRAs being used
-  - `Flash Attention` click to use flash attention in the model (if `Use extras` is enabled)
 
 </details>
 
@@ -566,7 +564,6 @@ If the user's GPU is not supported, the LLM will fall back to the CPU
 
 - `Show/Hide Advanced Options` Toggle to show/hide advanced options from below
 - `Log Level` select how verbose the log messages are
-- `Use extras` select to install and allow the use of extra features (flash attention and IQ quants)
 
 #### 💻 Setup Settings
 <div>

Runtime/LLM.cs

@@ -61,8 +61,6 @@ public class LLM : MonoBehaviour
         [ModelAdvanced] public string lora = "";
         /// <summary> the weights of the LORA models being used.</summary>
         [ModelAdvanced] public string loraWeights = "";
-        /// <summary> enable use of flash attention </summary>
-        [ModelExtras] public bool flashAttention = false;
 
         /// <summary> API key to use for the server (optional) </summary>
         public string APIKey;
@@ -435,7 +433,6 @@ protected virtual string GetLlamaccpArguments()
             if (numThreadsToUse > 0) arguments += $" -t {numThreadsToUse}";
             arguments += loraArgument;
             arguments += $" -ngl {numGPULayers}";
-            if (LLMUnitySetup.FullLlamaLib && flashAttention) arguments += $" --flash-attn";
             if (remote)
             {
                 arguments += $" --port {port} --host 0.0.0.0";

Runtime/LLMBuilder.cs

@@ -161,8 +161,6 @@ public static void HideLibraryPlatforms(string platform)
                 foreach (string platformPrefix in platforms)
                 {
                     bool move = sourceName.StartsWith(platformPrefix);
-                    move = move || (sourceName.Contains("cuda") && !sourceName.Contains("full") && LLMUnitySetup.FullLlamaLib);
-                    move = move || (sourceName.Contains("cuda") && sourceName.Contains("full") && !LLMUnitySetup.FullLlamaLib);
                     if (move)
                     {
                         string target = Path.Combine(BuildTempDir, sourceName);

Runtime/LLMLib.cs

@@ -462,18 +462,9 @@ public static List<string> PossibleArchitectures(bool gpu = false)
             {
                 if (gpu)
                 {
-                    if (LLMUnitySetup.FullLlamaLib)
-                    {
-                        architectures.Add("cuda-cu12.2.0-full");
-                        architectures.Add("cuda-cu11.7.1-full");
-                        architectures.Add("hip-full");
-                    }
-                    else
-                    {
-                        architectures.Add("cuda-cu12.2.0");
-                        architectures.Add("cuda-cu11.7.1");
-                        architectures.Add("hip");
-                    }
+                    architectures.Add("cuda-cu12.2.0");
+                    architectures.Add("cuda-cu11.7.1");
+                    architectures.Add("hip");
                     architectures.Add("vulkan");
                 }
                 if (has_avx512) architectures.Add("avx512");

Runtime/LLMUnitySetup.cs

@@ -59,7 +59,6 @@ public class LocalRemoteAttribute : PropertyAttribute {}
     public class RemoteAttribute : PropertyAttribute {}
     public class LocalAttribute : PropertyAttribute {}
     public class ModelAttribute : PropertyAttribute {}
-    public class ModelExtrasAttribute : PropertyAttribute {}
     public class ChatAttribute : PropertyAttribute {}
     public class LLMUnityAttribute : PropertyAttribute {}
 
@@ -112,8 +111,6 @@ public class LLMUnitySetup
         public static string libraryPath = GetAssetPath(libraryName);
         /// <summary> LlamaLib url </summary>
         public static string LlamaLibURL = $"{LlamaLibReleaseURL}/{libraryName}.zip";
-        /// <summary> LlamaLib extension url </summary>
-        public static string LlamaLibExtensionURL = $"{LlamaLibReleaseURL}/{libraryName}-full.zip";
         /// <summary> LLMnity store path </summary>
         public static string LLMUnityStore = Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.ApplicationData), "LLMUnity");
         /// <summary> Model download path </summary>
@@ -153,8 +150,6 @@ public class LLMUnitySetup
         /// \cond HIDE
         [LLMUnity] public static DebugModeType DebugMode = DebugModeType.All;
         static string DebugModeKey = "DebugMode";
-        public static bool FullLlamaLib = false;
-        static string FullLlamaLibKey = "FullLlamaLib";
         static List<Callback<string>> errorCallbacks = new List<Callback<string>>();
         static readonly object lockObject = new object();
         static Dictionary<string, Task> androidExtractTasks = new Dictionary<string, Task>();
@@ -189,7 +184,6 @@ public static void LogError(string message)
         static void LoadPlayerPrefs()
         {
             DebugMode = (DebugModeType)PlayerPrefs.GetInt(DebugModeKey, (int)DebugModeType.All);
-            FullLlamaLib = PlayerPrefs.GetInt(FullLlamaLibKey, 0) == 1;
         }
 
         public static void SetDebugMode(DebugModeType newDebugMode)
@@ -200,18 +194,6 @@ public static void SetDebugMode(DebugModeType newDebugMode)
             PlayerPrefs.Save();
         }
 
-#if UNITY_EDITOR
-        public static void SetFullLlamaLib(bool value)
-        {
-            if (FullLlamaLib == value) return;
-            FullLlamaLib = value;
-            PlayerPrefs.SetInt(FullLlamaLibKey, value ? 1 : 0);
-            PlayerPrefs.Save();
-            _ = DownloadLibrary();
-        }
-
-#endif
-
         public static string GetLibraryName(string version)
         {
             return $"undreamai-{version}-llamacpp";
@@ -452,9 +434,6 @@ static async Task DownloadLibrary()
 
                 // setup LlamaLib in StreamingAssets
                 await DownloadAndExtractInsideDirectory(LlamaLibURL, libraryPath, setupDir);
-
-                // setup LlamaLib extras in StreamingAssets
-                if (FullLlamaLib) await DownloadAndExtractInsideDirectory(LlamaLibExtensionURL, libraryPath, setupDir);
             }
             catch (Exception e)
             {