feat(workspace): prepare for the merge with ullm

.github/workflows/main.yml

@@ -15,10 +15,9 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
-      - uses: dtolnay/rust-toolchain@stable
-        with:
-          targets: wasm32-unknown-unknown
-      - uses: Swatinem/rust-cache@v2
+      - name: Install Toolchain
+        run: rustup update stable && rustup default stable
+      - uses: taiki-e/install-action@sccache
       - uses: taiki-e/install-action@nextest
 
       - name: Run Tests
@@ -29,9 +28,11 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
-      - uses: arduino/setup-protoc@v3
-      - uses: dtolnay/rust-toolchain@stable
-      - uses: Swatinem/rust-cache@v2
+      - name: Install Toolchain
+        run: |
+          rustup update stable && rustup default stable
+          rustup component add clippy rustfmt
+      - uses: taiki-e/install-action@sccache
       - name: Format
         run: cargo fmt --check
       - name: Clippy

Cargo.toml

@@ -1,25 +1,42 @@
 [workspace]
-resolver = "2"
-members = ["crates/*"]
+resolver = "3"
+members = ["legacy/*"]
+
+[workspace.package]
+version = "0.0.9"
+edition = "2024"
+authors = ["clearloop <[email protected]>"]
+license = "MIT"
+repository = "https://github.com/clearloop/cydonia"
 
 [workspace.dependencies]
-ccore = { path = "crates/core", package = "cydonia-core" }
+model = { path = "legacy/model", package = "cydonia-model" }
 candle = { path = "crates/candle", package = "cydonia-candle" }
 
 # crates.io
 anyhow = "1"
+async-stream = "0.3"
+bytes = "1.11.0"
+clap = { version = "4.5", features = ["derive"] }
+derive_more = { version = "2.1", features = ["display", "try_from"] }
+dirs = "6"
+futures-core = "0.3"
+futures-util = "0.3"
+reqwest = { version = "0.12", features = ["json", "stream"] }
+schemars = "1.1.0"
+serde = { version = "1", features = ["derive"] }
+serde_json = "1"
+tokio = { version = "1", features = ["rt-multi-thread", "macros"] }
+toml = "0.9.8"
+tracing = "0.1"
+tracing-subscriber = { version = "0.3", features = ["env-filter"] }
+
+# legacy dependencies
 candle-core = "0.8.1"
 candle-nn = "0.8.1"
 candle-transformers = "0.8.1"
-clap = "4.5.23"
-dirs = "5.0.1"
 hf-hub = "0.4.1"
 llamac-sys = { version = "0.1.86", package = "llama-cpp-sys-2" }
-once_cell = "1.20.2"
+once_cell = "1.21"
 rand = "0.8.5"
-serde = "1.0.217"
-serde_json = "1.0.134"
 tokenizers = "0.21.0"
-toml = "0.8.19"
-tracing = "0.1.41"
-tracing-subscriber = "0.3.19"

README.md

@@ -1,33 +1,3 @@
 # Cydonia
 
-Cydonia is a library based on [candle][candle] for developing modern AI applications in rust.
-
-```rust
-use cydonia::Model;
-
-fn main() {
-    let model = Model::new("gemma2").tag("latest");
-    let response = model.invoke("Hello, world!");
-    println!("{}", response);
-}
-```
-
-We support quantized models only derived from `gemma` and `llama` family.
-
-## TODOs
-
-- [x] Support chat interface ( history prompts )
-- [ ] Function encoder for llama3 tools (static)
-- [ ] Cydonia as service
-  - [ ] RPC support for llama3 tools (remote)
-  - [ ] GraphQL support for llama3 tools (remote)
-- [ ] RAG support
-- [ ] Agent interface
-- [ ] Multi-agent support (single-node)
-- [ ] An application based on the tools
-- [ ] p2p for the decentralized cydonia network (multi-node)
-- [ ] Test gpu
-
-<!-- links -->
-
-[candle]: https://github.com/huggingface/candle
+The Agent framework.

crates/candle/Cargo.toml → legacy/candle/Cargo.toml

@@ -9,7 +9,7 @@ anyhow.workspace = true
 candle-core.workspace = true
 candle-nn.workspace = true
 candle-transformers.workspace = true
-ccore.workspace = true
+model.workspace = true
 hf-hub.workspace = true
 rand.workspace = true
 serde.workspace = true

crates/candle/examples/llama.rs → legacy/candle/examples/llama.rs

@@ -1,5 +1,5 @@
-use ccore::{Message, Release};
 use cydonia_candle::{Llama, ProcessorConfig};
+use model::{Message, Release};
 use std::io::Write;
 
 fn main() {

crates/candle/src/inference.rs → legacy/candle/src/inference.rs

@@ -2,7 +2,7 @@
 use anyhow::Result;
 use candle_core::{quantized::gguf_file::Content, Device, Tensor};
 use candle_transformers::models::quantized_llama;
-use ccore::{chat, Message};
+use model::{chat, Message};
 use std::fs::File;
 
 /// The inference interface for language models

crates/candle/src/loader.rs → legacy/candle/src/loader.rs

@@ -3,8 +3,8 @@
 use crate::{Inference, Tokenizer};
 use anyhow::Result;
 use candle_core::Device;
-use ccore::{Release, TOKENIZER};
 use hf_hub::api::sync::Api;
+use model::{Release, TOKENIZER};
 use std::fs::File;
 
 /// Huggingface model loader

crates/candle/src/model.rs → legacy/candle/src/model.rs

@@ -2,7 +2,7 @@
 
 use crate::{Inference, Loader, Processor, ProcessorConfig, TokenStream, Tokenizer};
 use anyhow::Result;
-use ccore::{Message, Release};
+use model::{Message, Release};
 
 /// Language Model interface
 pub struct Model<I: Inference> {

crates/core/Cargo.toml → legacy/model/Cargo.toml

@@ -1,5 +1,5 @@
 [package]
-name = "cydonia-core"
+name = "cydonia-model"
 version = "0.0.0"
 edition = "2021"
 

legacy/model/README.md

@@ -0,0 +1 @@
+This crate is currently outdated.