chore: update rust versions to v0.2.0 (#155)

Co-authored-by: Ryan McCormick <rmccormick@nvidia.com>

.github/workflows/pre-merge-rust.yml

@@ -48,7 +48,7 @@ jobs:
       uses: actions/cache@v3
       with:
         path: ~/.cargo/bin
-        key: ${{ runner.os }}-cargo-tools-${{ hashFiles('**/Cargo.lock') }}
+        key: ${{ runner.os }}-cargo-tools-${{ github.head_ref || github.ref_name }}-${{ hashFiles('**/Cargo.lock') }}
     - name: Set up Rust Toolchain Components
       run: rustup component add rustfmt clippy
     - name: Run Cargo Check

llm/rust/Cargo.lock

 # This file is automatically @generated by Cargo.
 # It is not intended for manual editing.
-version = 3
+version = 4
 
 [[package]]
 name = "addr2line"
@@ -2839,7 +2839,7 @@ dependencies = [
 
 [[package]]
 name = "triton-distributed"
-version = "0.1.3"
+version = "0.2.0"
 dependencies = [
  "anyhow",
  "async-nats",
@@ -2877,7 +2877,7 @@ dependencies = [
 
 [[package]]
 name = "triton-llm"
-version = "0.1.3"
+version = "0.2.0"
 dependencies = [
  "anyhow",
  "async-stream",

llm/rust/Cargo.toml

@@ -20,7 +20,7 @@ members = [
 resolver = "2"
 
 [workspace.package]
-version = "0.1.3"
+version = "0.2.0"
 edition = "2021"
 authors = ["NVIDIA"]
 license = "Apache-2.0"
@@ -30,7 +30,7 @@ repository = "https://github.com/triton-inference-server/triton_distributed"
 
 [workspace.dependencies]
 # local or crates.io
-triton-distributed = { version = "0.1.3", path = "../../runtime/rust" }
+triton-distributed = { version = "0.2.0", path = "../../runtime/rust" }
 
 # crates.io
 anyhow = { version = "1" }

runtime/rust/Cargo.lock

@@ -2417,7 +2417,7 @@ dependencies = [
 
 [[package]]
 name = "triton-distributed"
-version = "0.1.3"
+version = "0.2.0"
 dependencies = [
  "anyhow",
  "assert_matches",

runtime/rust/Cargo.toml

@@ -17,7 +17,7 @@
 name = "triton-distributed"
 description = "Distributed GenAI Inference Framework"
 readme = "README.md"
-version = "0.1.3" # TODO: Centralize Version Automation
+version = "0.2.0" # TODO: Centralize Version Automation
 edition = "2021"
 authors = ["NVIDIA"]
 license = "Apache-2.0"

runtime/rust/python-wheel/Cargo.lock

@@ -2614,7 +2614,7 @@ dependencies = [
 
 [[package]]
 name = "triton-distributed"
-version = "0.1.3"
+version = "0.2.0"
 dependencies = [
  "anyhow",
  "async-nats",
@@ -2652,7 +2652,7 @@ dependencies = [
 
 [[package]]
 name = "triton_distributed_py3"
-version = "0.1.3"
+version = "0.2.0"
 dependencies = [
  "futures",
  "once_cell",

runtime/rust/python-wheel/Cargo.toml

@@ -15,7 +15,7 @@
 
 [package]
 name = "triton_distributed_py3"
-version = "0.1.3"
+version = "0.2.0"
 edition = "2021"
 authors = ["NVIDIA"]
 license = "Apache-2.0"
@@ -30,7 +30,7 @@ crate-type = ["cdylib"]
 
 
 [dependencies]
-triton-distributed = { version = "0.1.3", path = "../" }
+triton-distributed = { version = "0.2.0", path = "../" }
 
 futures = "0.3"
 once_cell = "1.20.3"

runtime/rust/python-wheel/pyproject.toml

@@ -16,7 +16,7 @@
 
 [project]
 name = "triton-distributed-rs"
-version = "0.1.3"
+version = "0.2.0"
 description = "Distributed LLM Framework"
 # readme = "README.md"
 authors = [