feature: init cli, upgrade llama-node with latest ggml implementation (#10)

* fix: download incorrect artifacts branch

* feat: impl cli

* fix: cli

* feat: upgrade to newest ggml

* feat: init cli

* chore: version

* fix: ci

* fix: ci

* chore: add cache to npm

Author: hlhr202

.github/workflows/llama-build.yml

@@ -44,13 +44,15 @@ jobs:
       - uses: actions/setup-node@v3
         with:
           node-version: '18'
+          cache: 'npm'
       # - name: Install ziglang
       #   uses: goto-bus-stop/setup-zig@v1
       #   with:
       #     version: 0.10.0
       # - run: cargo install cargo-xwin
       - run: npm install
       - run: cargo clippy --all-targets --all-features
+        working-directory: ./packages/core
       - run: npm run --workspace=packages/core cross-compile
       - run: npm run type-check
       - uses: actions/upload-artifact@v3

.gitignore

@@ -1,4 +1,5 @@
 target/
 node_modules/
 *.bin
-dist/
+dist/
+miscellaneous/models/7B

.npmignore

@@ -3,4 +3,5 @@ __test__/
 node_modules/
 target/
 *.bin
-rustfmt.toml
+rustfmt.toml
+miscellaneous/

.vscode/settings.json

@@ -0,0 +1,5 @@
+{
+    "rust-analyzer.linkedProjects": [
+        "packages/core/Cargo.toml",
+    ]
+}

Cargo.toml

@@ -249,3 +249,4 @@ llama
 - [ ] prompt extensions
 - [ ] more platforms and cross compile (performance related)
 - [ ] tweak embedding API, make end token configurable
+- [ ] cli and interactive

README.md

@@ -0,0 +1 @@
+eeec4125e9c7560836b4873b6f8e3025  tokenizer.model