fix: use a compressed logo image for README.md (#340)

* fix: use a compressed logo image for `README.md`
* build: switch to `latest` release of `llama.cpp`

Author: giladgd

.config/typedoc.css

@@ -215,6 +215,7 @@ h6:not(.tsd-anchor-link, .tsd-returns-title) > a:hover:before {
 }
 
 img[src$="assets/logo.v3.roundEdges.png"],
+img[src$="assets/logo.v3.roundEdges.avif"],
 img[src$="assets/logo.v3.png"] {
     box-shadow: 0px 4px 12px 0px rgb(0 0 0 / 16%), 0px 8px 64px 0px rgb(0 0 0 / 24%);
     border-radius: 14px;

.github/workflows/build.yml

@@ -23,8 +23,7 @@ jobs:
       - name: Download latest llama.cpp release
         env:
           CI: true
-        # Switched to `b3808` instead of `latest` due to a build failure on the latest version. `b3808` is the previous release.
-        run: node ./dist/cli/cli.js source download --release b3808 --skipBuild --noBundle --noUsageExample --updateBinariesReleaseMetadataAndSaveGitBundle
+        run: node ./dist/cli/cli.js source download --release latest --skipBuild --noBundle --noUsageExample --updateBinariesReleaseMetadataAndSaveGitBundle
       - name: Upload build artifact
         uses: actions/upload-artifact@v4
         with:
@@ -800,3 +799,35 @@ jobs:
         run: |
           curl -X POST "https://pubsubhubbub.appspot.com/" -H "Content-Type: application/x-www-form-urlencoded" --data-urlencode "hub.mode=publish" --data-urlencode "hub.url=https://node-llama-cpp.withcat.ai/blog/feed.atom"
 
+#  fix-broken-github-release:
+#    name: Fix broken GitHub release
+#    if: github.event_name == 'pull_request' && github.head_ref == 'gilad/smallerLogoFile'
+#    runs-on: ubuntu-latest
+#    permissions:
+#      id-token: write
+#      actions: read
+#      contents: write
+#      issues: write
+#      pull-requests: write
+#      discussions: write
+#    steps:
+#      - uses: actions/checkout@v4
+#      - uses: actions/setup-node@v4
+#        with:
+#          node-version: "20"
+#      - name: Install modules
+#        run: npm ci
+#
+#      - name: Pull artifact from broken release
+#        uses: actions/download-artifact@v4
+#        with:
+#          name: resolved-next-release
+#          github-token: ${{ secrets.GITHUB_TOKEN }}
+#          repository: "${{ github.repository }}"
+#          run-id: "<run id here>"
+#          path: scripts/resolved-next-release-artifact
+#
+#      - name: Fix broken release
+#        env:
+#          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+#        run: npx --no vite-node ./scripts/rerunSemanticReleaseGithubSuccess.ts

.vitepress/config.ts

@@ -33,7 +33,6 @@ const urlBase = env.get("DOCS_URL_BASE")
 const packageVersion = env.get("DOCS_PACKAGE_VERSION")
     .default(packageJson.version)
     .asString();
-const googleSiteVerificationCode = "7b4Hd_giIK0EFsin6a7PWLmM_OeaC7APLZUxVGwwI6Y";
 
 const hostname = "https://node-llama-cpp.withcat.ai/";
 
@@ -124,7 +123,6 @@ export default defineConfig({
     ],
     async transformHead({pageData, head}) {
         if (pageData.filePath === "index.md") {
-            head.push(["meta", {name: "google-site-verification", content: googleSiteVerificationCode}]);
             head.push(...defaultImageMetaTags);
         } else if (pageData.relativePath === "404.md")
             head.push(...defaultImageMetaTags);

.vitepress/theme/style.css

@@ -182,7 +182,8 @@ a.inlineCodeLink pre>code {
 img[src$="assets/logo.roundEdges.png"],
 img[src$="assets/logo.png"],
 img[src$="assets/logo.v3.roundEdges.png"],
-img[src$="assets/logo.v3.png"]{
+img[src$="assets/logo.v3.roundEdges.avif"],
+img[src$="assets/logo.v3.png"] {
     box-shadow: 0px 4px 12px 0px rgb(0 0 0 / 16%), 0px 8px 64px 0px rgb(0 0 0 / 24%);
     border-radius: 14px;
     margin-bottom: 12px;

README.md

@@ -1,5 +1,5 @@
 <div align="center">
-    <img alt="node-llama-cpp Logo" src="https://raw.githubusercontent.com/withcatai/node-llama-cpp/master/assets/logo.v3.roundEdges.png" width="360px" />
+    <img alt="node-llama-cpp Logo" src="https://raw.githubusercontent.com/withcatai/node-llama-cpp/master/assets/logo.v3.roundEdges.avif" width="360px" />
     <h1>node-llama-cpp</h1>
     <p>Run AI models locally on your machine</p>
     <sub>Pre-built bindings are provided with a fallback to building from source with cmake</sub>

assets/logo.v3.roundEdges.avif

@@ -1 +1,2 @@
+User-agent: *
 Sitemap: https://node-llama-cpp.withcat.ai/sitemap.xml

docs/public/robots.txt

@@ -0,0 +1,38 @@
+import path from "path";
+import {fileURLToPath} from "url";
+import fs from "fs-extra";
+// @ts-ignore
+import {verifyConditions as githubVerifyConditions, success as githubSuccess} from "@semantic-release/github";
+
+const __dirname = path.dirname(fileURLToPath(import.meta.url));
+
+const brokenReleaseDryRunResult = fs.readJSONSync(path.join(__dirname, "resolved-next-release-artifact", "semanticReleaseDryRunReleaseResult.json"));
+
+console.log("Broken release dry run result:", brokenReleaseDryRunResult);
+
+const githubPluginConfig = {};
+const context = {
+    ...brokenReleaseDryRunResult,
+    options: {
+        repositoryUrl: "https://github.com/withcatai/node-llama-cpp.git"
+    },
+    logger: console,
+    env: process.env
+};
+
+for (const release of context.releases) {
+    if (release.pluginName === "@semantic-release/npm") {
+        if (release.url == null) {
+            release.name = "npm package (@latest dist-tag)";
+            release.url = "https://www.npmjs.com/package/node-llama-cpp/v/" + release.version;
+        }
+    } else if (release.pluginName === "@semantic-release/github") {
+        if (release.url == null) {
+            release.name = "GitHub release";
+            release.url = "https://github.com/withcatai/node-llama-cpp/releases/tag/" + release.gitTag;
+        }
+    }
+}
+
+await githubVerifyConditions(githubPluginConfig, context);
+await githubSuccess(githubPluginConfig, context);

scripts/rerunSemanticReleaseGithubSuccess.ts

@@ -78,80 +78,93 @@ export const recommendedModels: ModelRecommendation[] = [{
         }
     }]
 }, {
-    name: "Phi 3 3.8B",
+    name: "Mistral Nemo 12B",
     abilities: ["chat", "complete", "functionCalling"],
-    description: "Phi 3 model was created by Microsoft and is optimized for strong reasoning (especially math and logic).\n" +
-        "This is the smallversion of the model.",
+    description: "Mistral Nemo model was created by Mistral AI and was trained on large proportion of multilingual and code data, with support for function calling.\n" +
+        "It was trained jointly by Mistral AI and NVIDIA.\n" +
+        "This is a 12 billion parameters model.",
 
     fileOptions: [{
         huggingFace: {
-            model: "bartowski/Phi-3.1-mini-4k-instruct-GGUF",
+            model: "mradermacher/Mistral-Nemo-Instruct-2407-GGUF",
             branch: "main",
-            file: "Phi-3.1-mini-4k-instruct-Q8_0.gguf"
+            file: "Mistral-Nemo-Instruct-2407.Q8_0.gguf"
         }
     }, {
         huggingFace: {
-            model: "bartowski/Phi-3.1-mini-4k-instruct-GGUF",
+            model: "mradermacher/Mistral-Nemo-Instruct-2407-GGUF",
             branch: "main",
-            file: "Phi-3.1-mini-4k-instruct-Q4_K_M.gguf"
+            file: "Mistral-Nemo-Instruct-2407.Q6_K.gguf"
         }
-    }]
-}, {
-    name: "Llama 2 Chat 7B",
-    abilities: ["chat", "complete"],
-    description: "Llama 2 Chat model was created by Meta and is optimized for an assistant-like chat use cases.\n" +
-        "This is the 7 billion parameters version of the model.",
-
-    fileOptions: [{
+    }, {
         huggingFace: {
-            model: "TheBloke/Llama-2-7B-Chat-GGUF",
+            model: "mradermacher/Mistral-Nemo-Instruct-2407-GGUF",
             branch: "main",
-            file: "llama-2-7b-chat.Q5_K_M.gguf"
+            file: "Mistral-Nemo-Instruct-2407.Q4_K_M.gguf"
         }
     }, {
         huggingFace: {
-            model: "TheBloke/Llama-2-7B-Chat-GGUF",
+            model: "mradermacher/Mistral-Nemo-Instruct-2407-GGUF",
             branch: "main",
-            file: "llama-2-7b-chat.Q4_K_M.gguf"
+            file: "Mistral-Nemo-Instruct-2407.Q4_K_S.gguf"
         }
     }]
 }, {
-    name: "Llama 2 Chat 13B",
-    abilities: ["chat", "complete"],
-    description: "Llama 2 Chat model was created by Meta and is optimized for an assistant-like chat use cases.\n" +
-        "This is the 13 billion parameters version of the model.",
+    name: "Phi 3 3.8B",
+    abilities: ["chat", "complete", "functionCalling"],
+    description: "Phi 3 model was created by Microsoft and is optimized for strong reasoning (especially math and logic).\n" +
+        "This is the smallversion of the model.",
 
     fileOptions: [{
         huggingFace: {
-            model: "TheBloke/Llama-2-13B-chat-GGUF",
+            model: "bartowski/Phi-3.1-mini-4k-instruct-GGUF",
             branch: "main",
-            file: "llama-2-13b-chat.Q5_K_M.gguf"
+            file: "Phi-3.1-mini-4k-instruct-Q8_0.gguf"
         }
     }, {
         huggingFace: {
-            model: "TheBloke/Llama-2-13B-chat-GGUF",
+            model: "bartowski/Phi-3.1-mini-4k-instruct-GGUF",
             branch: "main",
-            file: "llama-2-13b-chat.Q4_K_M.gguf"
+            file: "Phi-3.1-mini-4k-instruct-Q4_K_M.gguf"
         }
     }]
 }, {
-    name: "Llama 2 Chat 70B",
-    abilities: ["chat", "complete"],
-    description: "Llama 2 Chat model was created by Meta and is optimized for an assistant-like chat use cases.\n" +
-        "This is the 70 billion parameters version of the model. " +
-        "You need a GPU with a lot of VRAM to use this version.",
+    name: "OLMoE 1B 7B MoE",
+    abilities: ["chat"],
+    description: "OLMoE models were created by AllenAI, and are fully open source models that utilize a Mixture of Experts architecture.\n" +
+        "Mixtures of Experts (MoE) is a technique where different models, each skilled in solving a particular kind of problem, work together to the improve the overall performance on complex tasks.\n" +
+        "This model includes 64 expert models, with a total of 7 billion parameters.\n" +
+        "This model generates output extremely fast.",
 
     fileOptions: [{
         huggingFace: {
-            model: "TheBloke/Llama-2-70B-Chat-GGUF",
+            model: "allenai/OLMoE-1B-7B-0924-Instruct-GGUF",
             branch: "main",
-            file: "llama-2-70b-chat.Q5_K_M.gguf"
+            file: "olmoe-1b-7b-0924-instruct-q8_0.gguf"
         }
     }, {
         huggingFace: {
-            model: "TheBloke/Llama-2-70B-Chat-GGUF",
+            model: "allenai/OLMoE-1B-7B-0924-Instruct-GGUF",
             branch: "main",
-            file: "llama-2-70b-chat.Q4_K_M.gguf"
+            file: "olmoe-1b-7b-0924-instruct-q6_k.gguf"
+        }
+    }, {
+        huggingFace: {
+            model: "allenai/OLMoE-1B-7B-0924-Instruct-GGUF",
+            branch: "main",
+            file: "olmoe-1b-7b-0924-instruct-q5_k_m.gguf"
+        }
+    }, {
+        huggingFace: {
+            model: "allenai/OLMoE-1B-7B-0924-Instruct-GGUF",
+            branch: "main",
+            file: "olmoe-1b-7b-0924-instruct-q4_k_s.gguf"
+        }
+    }, {
+        huggingFace: {
+            model: "allenai/OLMoE-1B-7B-0924-Instruct-GGUF",
+            branch: "main",
+            file: "olmoe-1b-7b-0924-instruct-q4_k_m.gguf"
         }
     }]
 }, {
@@ -213,90 +226,6 @@ export const recommendedModels: ModelRecommendation[] = [{
             file: "dolphin-2.5-mixtral-8x7b.Q4_K_M.gguf"
         }
     }]
-}, /* {
-    name: "Functionary Medium v2.4",
-    abilities: ["chat", "complete", "functionCalling"],
-    description: "Functionary models were created by Meetkai and are optimized for function calling.\n" +
-        "This is the medium version of the model.",
-
-    fileOptions: [{
-        huggingFace: {
-            model: "meetkai/functionary-medium-v2.4-GGUF",
-            branch: "main",
-            file: "functionary-medium-v2.4.Q8_0.gguf"
-        }
-    }, {
-        huggingFace: {
-            model: "meetkai/functionary-medium-v2.4-GGUF",
-            branch: "main",
-            file: "functionary-medium-v2.4.Q4_0.gguf"
-        }
-    }]
-}, */ /* {
-    name: "Functionary Small v2.5",
-    abilities: ["chat", "complete", "functionCalling"],
-    description: "Functionary models were created by Meetkai and are optimized for function calling.\n" +
-        "This model is based on Llama 3.\n" +
-        "This is the small version of the model.",
-
-    fileOptions: [{
-        huggingFace: {
-            model: "meetkai/functionary-small-v2.5-GGUF",
-            branch: "main",
-            file: "functionary-small-v2.5.f16.gguf"
-        }
-    }, {
-        huggingFace: {
-            model: "meetkai/functionary-small-v2.5-GGUF",
-            branch: "main",
-            file: "functionary-small-v2.5.Q8_0.gguf"
-        }
-    }, {
-        huggingFace: {
-            model: "meetkai/functionary-small-v2.5-GGUF",
-            branch: "main",
-            file: "functionary-small-v2.5.Q4_0.gguf"
-        }
-    }]
-}, */ {
-    name: "OLMoE 1b 7B MoE",
-    abilities: ["chat"],
-    description: "OLMoE models were created by AllenAI, and are fully open source models that utilize a Mixture of Experts architecture" +
-        "Mixtures of Experts (MoE) is a technique where different models, each skilled in solving a particular kind of problem, work together to the improve the overall performance on complex tasks.\n" +
-        "This model includes 64 expert models, with a total of 7 billion parameters.\n" +
-        "This model generates output extremely fast.",
-
-    fileOptions: [{
-        huggingFace: {
-            model: "allenai/OLMoE-1B-7B-0924-Instruct-GGUF",
-            branch: "main",
-            file: "olmoe-1b-7b-0924-instruct-q8_0.gguf"
-        }
-    }, {
-        huggingFace: {
-            model: "allenai/OLMoE-1B-7B-0924-Instruct-GGUF",
-            branch: "main",
-            file: "olmoe-1b-7b-0924-instruct-q6_k.gguf"
-        }
-    }, {
-        huggingFace: {
-            model: "allenai/OLMoE-1B-7B-0924-Instruct-GGUF",
-            branch: "main",
-            file: "olmoe-1b-7b-0924-instruct-q5_k_m.gguf"
-        }
-    }, {
-        huggingFace: {
-            model: "allenai/OLMoE-1B-7B-0924-Instruct-GGUF",
-            branch: "main",
-            file: "olmoe-1b-7b-0924-instruct-q4_k_s.gguf"
-        }
-    }, {
-        huggingFace: {
-            model: "allenai/OLMoE-1B-7B-0924-Instruct-GGUF",
-            branch: "main",
-            file: "olmoe-1b-7b-0924-instruct-q4_k_m.gguf"
-        }
-    }]
 }, {
     name: "Gemma 2 9B",
     abilities: ["chat", "complete"],