Cuda and Cblast binaries

Author: SignalRT

.github/workflows/compile.yml

@@ -175,6 +175,13 @@ jobs:
             .\build\bin\Release\llama.dll
             .\build\bin\Release\clblast.dll
           name: llama-bin-win-clblast-x64.dll
+      - name: Upload llava artifacts (Windows)
+        if: ${{ matrix.os == 'windows-latest' }}
+        uses: actions/upload-artifact@v4
+        with:
+          path: |
+            .\build\bin\Release\llava_shared.dll
+          name: llava-bin-win-clblast-x64.dll
       - name: Upload artifacts (linux)
         if: ${{ matrix.os == 'ubuntu-22.04' }}
         uses: actions/upload-artifact@v4
@@ -183,6 +190,14 @@ jobs:
             ./build/libllama.so
             # ./build/libclblast.so
           name: llama-bin-linux-clblast-x64.so
+      - name: Upload llava artifacts (linux)
+        if: ${{ matrix.os == 'ubuntu-22.04' }}
+        uses: actions/upload-artifact@v4
+        with:
+          path: |
+            ./build/examples/llava/libllava_shared.so
+          name: llava-bin-linux-clblast-x64.so
+
 
   compile-cublas:
     name: Compile (cublas)
@@ -232,12 +247,25 @@ jobs:
         with:
           path: .\build\bin\Release\llama.dll
           name: llama-bin-win-cublas-cu${{ matrix.cuda }}-x64.dll
+      - name: Upload llava artifacts (Windows)
+        if: ${{ matrix.os == 'windows-latest' }}
+        uses: actions/upload-artifact@v4
+        with:
+          path: .\build\bin\Release\llava_shared.dll
+          name: llava-bin-win-cublas-cu${{ matrix.cuda }}-x64.dll
       - name: Upload artifacts (Linux)
         if: ${{ matrix.os == 'ubuntu-20.04' }}
         uses: actions/upload-artifact@v4
         with:
           path: ./build/libllama.so
           name: llama-bin-linux-cublas-cu${{ matrix.cuda }}-x64.so
+      - name: Upload llava artifacts (Linux)
+        if: ${{ matrix.os == 'ubuntu-20.04' }}
+        uses: actions/upload-artifact@v4
+        with:
+          path: ./build/examples/llava/libllava_shared.so
+          name: llava-bin-linux-cublas-cu${{ matrix.cuda }}-x64.so
+ 
 
   compile-macos:
     name: Compile (MacOS)

.github/workflows/main.yml

@@ -1,9 +1,9 @@
 name: CI
 on:
   push:
-    branches: [master,LlavaAPI]
-  pull_request:
-    branches: [master]
+    branches: [LlavaAPI]
+    # pull_request:
+    # branches: [master]
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}
   cancel-in-progress: true

LLama.Unittest/LLavaWeightsTests.cs

@@ -32,69 +32,7 @@ public void Dispose()
             _lLavaWeights.Dispose();
         }
 
-        // [Fact]
-        // public void CheckProperties()
-        // {
-        //     Assert.Equal(4096, (int)_context.ContextSize);
-        //     Assert.Equal(4096, _context.EmbeddingSize);
-        //     Assert.Equal(32000, _context.VocabCount);
-        // }
-
-        /*[Fact]
-        public void Tokenize()
-        {
-            var tokens = _context.Tokenize("The quick brown fox", true);
-
-            Assert.Equal(new LLamaToken[] { 1, 450, 4996, 17354, 1701, 29916 }, tokens);
-        }
-
-        [Fact]
-        public void TokenizeNewline()
-        {
-            var tokens = _context.Tokenize("\n", false, false);
-
-            Assert.Equal(new LLamaToken[] { 29871, 13 }, tokens);
-        }
-
-        [Fact]
-        public void TokenizeRoundtripSpecialStrings()
-        {
-            var strings = new[]
-            {
-                "\t", "\t\t", "\t\t\t",
-                "\n\n", "\n\n\n", "\n\n\n\n",
-                "\t\n", "\t\n\t\n\n\n\n\t\t",
-                "\b", "\v", "\0"
-            };
-
-            foreach (var s in strings)
-            {
-                var tokens = _context.Tokenize(s, false, false);
-                var decoder = new StreamingTokenDecoder(_context);
-                decoder.AddRange(tokens);
-
-                var str = decoder.Read();
-
-                Assert.Equal(s, str.TrimStart(' '));
-            }
-        }
-
-        [Fact]
-        public void TokenizeWithoutBOS()
-        {
-            var tokens = _context.Tokenize("The quick brown fox", false);
-
-            Assert.Equal(new LLamaToken[] { 450, 4996, 17354, 1701, 29916 }, tokens);
-        }
-
-        [Fact]
-        public void TokenizeEmpty()
-        {
-            var tokens = _context.Tokenize("", false);
-
-            Assert.Equal(Array.Empty<LLamaToken>(), tokens);
-        }*/
-        
+      
 
         [Fact]
         public void EmbedImageAsFileName()