chore: rename dynamo (#44)

Co-authored-by: Biswa Panda <biswa.panda@gmail.com>

launch/dynemo-run/src/input/http.rs → launch/dynamo-run/src/input/http.rs

@@ -15,7 +15,7 @@
 
 use std::sync::Arc;
 
-use dynemo_llm::{
+use dynamo_llm::{
     backend::Backend,
     http::service::{discovery, service_v2},
     model_type::ModelType,
@@ -27,7 +27,7 @@ use dynemo_llm::{
         Annotated,
     },
 };
-use dynemo_runtime::{
+use dynamo_runtime::{
     pipeline::{ManyOut, Operator, ServiceBackend, ServiceFrontend, SingleIn, Source},
     DistributedRuntime, Runtime,
 };

launch/dynemo-run/src/input/text.rs → launch/dynamo-run/src/input/text.rs

@@ -13,7 +13,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-use dynemo_llm::{
+use dynamo_llm::{
     backend::Backend,
     preprocessor::OpenAIPreprocessor,
     types::{
@@ -24,7 +24,7 @@ use dynemo_llm::{
         Annotated,
     },
 };
-use dynemo_runtime::{
+use dynamo_runtime::{
     pipeline::{Context, ManyOut, Operator, ServiceBackend, ServiceFrontend, SingleIn, Source},
     runtime::CancellationToken,
     DistributedRuntime, Runtime,

launch/dynemo-run/src/lib.rs → launch/dynamo-run/src/lib.rs

@@ -16,11 +16,11 @@
 #[cfg(any(feature = "vllm", feature = "sglang"))]
 use std::{future::Future, pin::Pin};
 
-use dynemo_llm::{
+use dynamo_llm::{
     backend::ExecutionContext, model_card::model::ModelDeploymentCard,
     types::openai::chat_completions::OpenAIChatCompletionsStreamingEngine,
 };
-use dynemo_runtime::protocols::Endpoint;
+use dynamo_runtime::protocols::Endpoint;
 
 mod flags;
 pub use flags::Flags;
@@ -67,7 +67,7 @@ pub enum EngineConfig {
 
 #[allow(unused_mut)]
 pub async fn run(
-    runtime: dynemo_runtime::Runtime,
+    runtime: dynamo_runtime::Runtime,
     mut in_opt: Input, // mut because vllm and sglang multi-node can change it
     out_opt: Output,
     flags: Flags,
@@ -152,12 +152,12 @@ pub async fn run(
             };
             EngineConfig::StaticFull {
                 service_name: model_name,
-                engine: dynemo_llm::engines::mistralrs::make_engine(&model_path).await?,
+                engine: dynamo_llm::engines::mistralrs::make_engine(&model_path).await?,
             }
         }
         #[cfg(feature = "sglang")]
         Output::SgLang => {
-            use dynemo_llm::engines::sglang;
+            use dynamo_llm::engines::sglang;
             let Some(model_path) = model_path else {
                 anyhow::bail!("out=sglang requires flag --model-path=<full-path-to-model-dir>");
             };
@@ -169,7 +169,7 @@ pub async fn run(
             let Some(sock_prefix) = zmq_socket_prefix else {
                 anyhow::bail!("sglang requires zmq_socket_prefix");
             };
-            let node_conf = dynemo_llm::engines::MultiNodeConfig {
+            let node_conf = dynamo_llm::engines::MultiNodeConfig {
                 num_nodes: flags.num_nodes,
                 node_rank: flags.node_rank,
                 leader_addr: flags.leader_addr.unwrap_or_default(),
@@ -207,7 +207,7 @@ pub async fn run(
         }
         #[cfg(feature = "vllm")]
         Output::Vllm => {
-            use dynemo_llm::engines::vllm;
+            use dynamo_llm::engines::vllm;
             if flags.base_gpu_id != 0 {
                 anyhow::bail!("vllm does not support base_gpu_id. Set environment variable CUDA_VISIBLE_DEVICES instead.");
             }
@@ -231,7 +231,7 @@ pub async fn run(
             let Some(sock_prefix) = zmq_socket_prefix else {
                 anyhow::bail!("vllm requires zmq_socket_prefix");
             };
-            let node_conf = dynemo_llm::engines::MultiNodeConfig {
+            let node_conf = dynamo_llm::engines::MultiNodeConfig {
                 num_nodes: flags.num_nodes,
                 node_rank: flags.node_rank,
                 leader_addr: flags.leader_addr.unwrap_or_default(),
@@ -274,7 +274,7 @@ pub async fn run(
         }
         #[cfg(feature = "llamacpp")]
         Output::LlamaCpp => {
-            use dynemo_llm::engines::llamacpp;
+            use dynamo_llm::engines::llamacpp;
             let Some(model_path) = model_path else {
                 anyhow::bail!("out=llamacpp requires flag --model-path=<full-path-to-model-gguf>");
             };
@@ -295,7 +295,7 @@ pub async fn run(
         }
         #[cfg(feature = "trtllm")]
         Output::TrtLLM => {
-            use dynemo_llm::engines::trtllm;
+            use dynamo_llm::engines::trtllm;
             let Some(model_path) = model_path else {
                 anyhow::bail!("out=trtllm requires flag --model-path=<full-path-to-model-dir>");
             };
@@ -315,7 +315,7 @@ pub async fn run(
         }
         #[cfg(feature = "python")]
         Output::PythonStr(path_str) => {
-            use dynemo_llm::engines::python;
+            use dynamo_llm::engines::python;
             let Some(model_name) = model_name else {
                 anyhow::bail!("Provide model service name as `--model-name <this>`");
             };
@@ -328,7 +328,7 @@ pub async fn run(
         }
         #[cfg(feature = "python")]
         Output::PythonTok(path_str) => {
-            use dynemo_llm::engines::python;
+            use dynamo_llm::engines::python;
             let Some(card) = maybe_card.clone() else {
                 anyhow::bail!("Could not find tokenizer. Pass flag --model-path <path>");
             };

launch/dynemo-run/src/main.rs → launch/dynamo-run/src/main.rs

@@ -17,17 +17,17 @@ use std::env;
 
 use clap::Parser;
 
-use dynemo_run::{Input, Output};
-use dynemo_runtime::logging;
+use dynamo_run::{Input, Output};
+use dynamo_runtime::logging;
 
 const HELP: &str = r#"
-dynemo-run is a single binary that wires together the various inputs (http, text, network) and workers (network, engine), that runs the services. It is the simplest way to use dynemo locally.
+dynamo-run is a single binary that wires together the various inputs (http, text, network) and workers (network, engine), that runs the services. It is the simplest way to use dynamo locally.
 
 Example:
 - cargo build --release --features mistralrs,cuda
 - cd target/release
-- ./dynemo-run hf_checkouts/Llama-3.2-3B-Instruct/
-- OR: ./dynemo-run Llama-3.2-1B-Instruct-Q4_K_M.gguf
+- ./dynamo-run hf_checkouts/Llama-3.2-3B-Instruct/
+- OR: ./dynamo-run Llama-3.2-1B-Instruct-Q4_K_M.gguf
 
 "#;
 
@@ -41,14 +41,14 @@ const DEFAULT_OUT: Output = Output::EchoFull;
 
 const ZMQ_SOCKET_PREFIX: &str = "dyn";
 
-const USAGE: &str = "USAGE: dynemo-run in=[http|text|dyn://<path>|none] out=[mistralrs|sglang|llamacpp|vllm|trtllm|echo_full|echo_core|pystr:<engine.py>|pytok:<engine.py>] [--http-port 8080] [--model-path <path>] [--model-name <served-model-name>] [--model-config <hf-repo>] [--tensor-parallel-size=1] [--num-nodes=1] [--node-rank=0] [--leader-addr=127.0.0.1:9876] [--base-gpu-id=0]";
+const USAGE: &str = "USAGE: dynamo-run in=[http|text|dyn://<path>|none] out=[mistralrs|sglang|llamacpp|vllm|trtllm|echo_full|echo_core|pystr:<engine.py>|pytok:<engine.py>] [--http-port 8080] [--model-path <path>] [--model-name <served-model-name>] [--model-config <hf-repo>] [--tensor-parallel-size=1] [--num-nodes=1] [--node-rank=0] [--leader-addr=127.0.0.1:9876] [--base-gpu-id=0]";
 
 fn main() -> anyhow::Result<()> {
     logging::init();
 
     // Call sub-processes before starting the Runtime machinery
     // For anything except sub-process starting try_parse_from will error.
-    if let Ok(flags) = dynemo_run::Flags::try_parse_from(env::args()) {
+    if let Ok(flags) = dynamo_run::Flags::try_parse_from(env::args()) {
         #[allow(unused_variables)]
         if let Some(sglang_flags) = flags.internal_sglang_process {
             let Some(model_path) = flags.model_path_flag.as_ref() else {
@@ -60,13 +60,13 @@ fn main() -> anyhow::Result<()> {
             if cfg!(feature = "sglang") {
                 #[cfg(feature = "sglang")]
                 {
-                    use dynemo_llm::engines::sglang;
+                    use dynamo_llm::engines::sglang;
                     let gpu_config = sglang::MultiGPUConfig {
                         tp_size: flags.tensor_parallel_size,
                         tp_rank: sglang_flags.tp_rank,
                         gpu_id: sglang_flags.gpu_id,
                     };
-                    let node_config = dynemo_llm::engines::MultiNodeConfig {
+                    let node_config = dynamo_llm::engines::MultiNodeConfig {
                         num_nodes: flags.num_nodes,
                         node_rank: flags.node_rank,
                         leader_addr: flags.leader_addr.unwrap_or_default(),
@@ -98,8 +98,8 @@ fn main() -> anyhow::Result<()> {
             if cfg!(feature = "vllm") {
                 #[cfg(feature = "vllm")]
                 {
-                    use dynemo_llm::engines::vllm;
-                    let node_config = dynemo_llm::engines::MultiNodeConfig {
+                    use dynamo_llm::engines::vllm;
+                    let node_config = dynamo_llm::engines::MultiNodeConfig {
                         num_nodes: flags.num_nodes,
                         node_rank: flags.node_rank,
                         leader_addr: flags.leader_addr.unwrap_or_default(),
@@ -119,15 +119,15 @@ fn main() -> anyhow::Result<()> {
     }
 
     // max_worker_threads and max_blocking_threads from env vars or config file.
-    let rt_config = dynemo_runtime::RuntimeConfig::from_settings()?;
+    let rt_config = dynamo_runtime::RuntimeConfig::from_settings()?;
 
     // One per process. Wraps a Runtime with holds two tokio runtimes.
-    let worker = dynemo_runtime::Worker::from_config(rt_config)?;
+    let worker = dynamo_runtime::Worker::from_config(rt_config)?;
 
     worker.execute(wrapper)
 }
 
-async fn wrapper(runtime: dynemo_runtime::Runtime) -> anyhow::Result<()> {
+async fn wrapper(runtime: dynamo_runtime::Runtime) -> anyhow::Result<()> {
     let mut in_opt = None;
     let mut out_opt = None;
     let args: Vec<String> = env::args().skip(1).collect();
@@ -171,13 +171,13 @@ async fn wrapper(runtime: dynemo_runtime::Runtime) -> anyhow::Result<()> {
 
     // Clap skips the first argument expecting it to be the binary name, so add it back
     // Note `--model-path` has index=1 (in lib.rs) so that doesn't need a flag.
-    let flags = dynemo_run::Flags::try_parse_from(
-        ["dynemo-run".to_string()]
+    let flags = dynamo_run::Flags::try_parse_from(
+        ["dynamo-run".to_string()]
             .into_iter()
             .chain(env::args().skip(non_flag_params)),
     )?;
 
-    dynemo_run::run(
+    dynamo_run::run(
         runtime,
         in_opt,
         out_opt,

launch/dynemo-run/src/output/echo_core.rs → launch/dynamo-run/src/output/echo_core.rs

@@ -18,12 +18,12 @@ use std::{sync::Arc, time::Duration};
 use async_stream::stream;
 use async_trait::async_trait;
 
-use dynemo_llm::backend::ExecutionContext;
-use dynemo_llm::preprocessor::BackendInput;
-use dynemo_llm::protocols::common::llm_backend::LLMEngineOutput;
-use dynemo_runtime::engine::{AsyncEngine, AsyncEngineContextProvider, ResponseStream};
-use dynemo_runtime::pipeline::{Error, ManyOut, SingleIn};
-use dynemo_runtime::protocols::annotated::Annotated;
+use dynamo_llm::backend::ExecutionContext;
+use dynamo_llm::preprocessor::BackendInput;
+use dynamo_llm::protocols::common::llm_backend::LLMEngineOutput;
+use dynamo_runtime::engine::{AsyncEngine, AsyncEngineContextProvider, ResponseStream};
+use dynamo_runtime::pipeline::{Error, ManyOut, SingleIn};
+use dynamo_runtime::protocols::annotated::Annotated;
 
 /// How long to sleep between echoed tokens.
 /// 50ms gives us 20 tok/s.

launch/dynemo-run/src/output/echo_full.rs → launch/dynamo-run/src/output/echo_full.rs

@@ -18,13 +18,13 @@ use std::{sync::Arc, time::Duration};
 use async_stream::stream;
 use async_trait::async_trait;
 
-use dynemo_llm::protocols::openai::chat_completions::{
+use dynamo_llm::protocols::openai::chat_completions::{
     NvCreateChatCompletionRequest, NvCreateChatCompletionStreamResponse,
 };
-use dynemo_llm::types::openai::chat_completions::OpenAIChatCompletionsStreamingEngine;
-use dynemo_runtime::engine::{AsyncEngine, AsyncEngineContextProvider, ResponseStream};
-use dynemo_runtime::pipeline::{Error, ManyOut, SingleIn};
-use dynemo_runtime::protocols::annotated::Annotated;
+use dynamo_llm::types::openai::chat_completions::OpenAIChatCompletionsStreamingEngine;
+use dynamo_runtime::engine::{AsyncEngine, AsyncEngineContextProvider, ResponseStream};
+use dynamo_runtime::pipeline::{Error, ManyOut, SingleIn};
+use dynamo_runtime::protocols::annotated::Annotated;
 
 /// How long to sleep between echoed tokens.
 /// 50ms gives us 20 tok/s.

lib/bindings/c/Cargo.lock

@@ -955,7 +955,7 @@ dependencies = [
 ]
 
 [[package]]
-name = "dynemo-llm"
+name = "dynamo-llm"
 version = "0.2.1"
 dependencies = [
  "anyhow",
@@ -970,7 +970,7 @@ dependencies = [
  "chrono",
  "cmake",
  "derive_builder",
- "dynemo-runtime",
+ "dynamo-runtime",
  "either",
  "erased-serde",
  "futures",
@@ -1004,7 +1004,7 @@ dependencies = [
 ]
 
 [[package]]
-name = "dynemo-runtime"
+name = "dynamo-runtime"
 version = "0.2.1"
 dependencies = [
  "anyhow",
@@ -1947,14 +1947,14 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b5aba8db14291edd000dfcc4d620c7ebfb122c613afb886ca8803fa4e128a20a"
 
 [[package]]
-name = "libdynemo-llm"
+name = "libdynamo_llm"
 version = "0.2.1"
 dependencies = [
  "anyhow",
  "async-once-cell",
  "cbindgen",
- "dynemo-llm",
- "dynemo-runtime",
+ "dynamo-llm",
+ "dynamo-runtime",
  "futures",
  "libc",
  "once_cell",

lib/bindings/c/Cargo.toml

@@ -14,7 +14,7 @@
 # limitations under the License.
 
 [package]
-name = "libdynemo-llm"
+name = "libdynamo_llm"
 version = "0.2.1"
 edition = "2021"
 authors = ["NVIDIA"]
@@ -23,15 +23,15 @@ homepage = "https://github.com/dynemo-ai/dynemo"
 repository = "https://github.com/dynemo-ai/dynemo.git"
 
 [lib]
-name = "dynemo_llm_capi"
+name = "dynamo_llm_capi"
 crate-type = ["cdylib"]
 
 [build-dependencies]
 cbindgen = "0.27"
 
 [dependencies]
-dynemo-llm = { path = "../../llm" }
-dynemo-runtime = { path = "../../runtime" }
+dynamo-llm = { path = "../../llm" }
+dynamo-runtime = { path = "../../runtime" }
 
 anyhow = { version = "1" }
 futures = "0.3"

lib/bindings/c/build.rs

@@ -22,7 +22,7 @@ fn main() {
     let header_path = Path::new(&crate_dir)
         .join("include")
         .join("nvidia")
-        .join("dynemo_llm")
+        .join("dynamo_llm")
         .join("llm_engine.h");
 
     cbindgen::generate(crate_dir)

lib/bindings/c/cbindgen.toml

@@ -25,7 +25,7 @@ enum_class = false
 
 
 [export]
-include = ["DynemoLlmResult", "dynemo_llm_init", "dynemo_llm_shutdown"]
+include = ["DynamoLlmResult", "dynamo_llm_init", "dynamo_llm_shutdown"]
 
 [export.rename]
-"DynemoLlmResult" = "dynemo_llm_result_t"
+"DynamoLlmResult" = "dynamo_llm_result_t"

lib/bindings/c/src/lib.rs

@@ -19,10 +19,10 @@ use once_cell::sync::OnceCell;
 use std::ffi::CStr;
 use std::sync::atomic::{AtomicU32, Ordering};
 
-use dynemo_llm::kv_router::{
+use dynamo_llm::kv_router::{
     indexer::compute_block_hash_for_seq, protocols::*, publisher::KvEventPublisher,
 };
-use dynemo_runtime::{DistributedRuntime, Worker};
+use dynamo_runtime::{DistributedRuntime, Worker};
 static WK: OnceCell<Worker> = OnceCell::new();
 static DRT: AsyncOnceCell<DistributedRuntime> = AsyncOnceCell::new();
 // [FIXME] shouldn't the publisher be instance passing between API calls?
@@ -41,7 +41,7 @@ fn initialize_tracing() {
 }
 
 #[repr(u32)]
-pub enum DynemoLlmResult {
+pub enum DynamoLlmResult {
     OK = 0,
     ERR = 1,
 }
@@ -49,17 +49,17 @@ pub enum DynemoLlmResult {
 /// # Safety
 /// the namespace_c_str and component_c_str are passed as pointers to C strings
 #[no_mangle]
-pub unsafe extern "C" fn dynemo_llm_init(
+pub unsafe extern "C" fn dynamo_llm_init(
     namespace_c_str: *const c_char,
     component_c_str: *const c_char,
     worker_id: i64,
-) -> DynemoLlmResult {
+) -> DynamoLlmResult {
     initialize_tracing();
     let wk = match WK.get_or_try_init(Worker::from_settings) {
         Ok(wk) => wk.clone(),
         Err(e) => {
             eprintln!("Failed to initialize runtime: {:?}", e);
-            return DynemoLlmResult::ERR;
+            return DynamoLlmResult::ERR;
         }
     };
     let rt = wk.runtime();
@@ -73,7 +73,7 @@ pub unsafe extern "C" fn dynemo_llm_init(
             Ok(_) => Ok(()),
             Err(e) => {
                 eprintln!("Failed to initialize distributed runtime: {:?}", e);
-                Err(DynemoLlmResult::ERR)
+                Err(DynamoLlmResult::ERR)
             }
         }
     });
@@ -81,7 +81,7 @@ pub unsafe extern "C" fn dynemo_llm_init(
         Ok(s) => s.to_string(),
         Err(e) => {
             eprintln!("Failed to convert C string to Rust string: {:?}", e);
-            return DynemoLlmResult::ERR;
+            return DynamoLlmResult::ERR;
         }
     };
 
@@ -89,18 +89,18 @@ pub unsafe extern "C" fn dynemo_llm_init(
         Ok(s) => s.to_string(),
         Err(e) => {
             eprintln!("Failed to convert C string to Rust string: {:?}", e);
-            return DynemoLlmResult::ERR;
+            return DynamoLlmResult::ERR;
         }
     };
 
     match result {
         Ok(_) => match KV_PUB
-            .get_or_try_init(move || dynemo_create_kv_publisher(namespace, component, worker_id))
+            .get_or_try_init(move || dynamo_create_kv_publisher(namespace, component, worker_id))
         {
-            Ok(_) => DynemoLlmResult::OK,
+            Ok(_) => DynamoLlmResult::OK,
             Err(e) => {
                 eprintln!("Failed to initialize distributed runtime: {:?}", e);
-                DynemoLlmResult::ERR
+                DynamoLlmResult::ERR
             }
         },
         Err(e) => e,
@@ -108,33 +108,33 @@ pub unsafe extern "C" fn dynemo_llm_init(
 }
 
 #[no_mangle]
-pub extern "C" fn dynemo_llm_shutdown() -> DynemoLlmResult {
+pub extern "C" fn dynamo_llm_shutdown() -> DynamoLlmResult {
     let wk = match WK.get() {
         Some(wk) => wk,
         None => {
             eprintln!("Runtime not initialized");
-            return DynemoLlmResult::ERR;
+            return DynamoLlmResult::ERR;
         }
     };
 
     wk.runtime().shutdown();
 
-    DynemoLlmResult::OK
+    DynamoLlmResult::OK
 }
 
 #[no_mangle]
-pub extern "C" fn dynemo_llm_load_publisher_create() -> DynemoLlmResult {
-    DynemoLlmResult::OK
+pub extern "C" fn dynamo_llm_load_publisher_create() -> DynamoLlmResult {
+    DynamoLlmResult::OK
 }
 
 // instantiate a kv publisher
 // this will bring up the task to publish and the channels to await publishing events
-// the [`dynemo_kv_publish_store_event`] call will use a handle to the publisher to send events
-// store and the [`dynemo_kv_event_create_removed`] will create remove events
+// the [`dynamo_kv_publish_store_event`] call will use a handle to the publisher to send events
+// store and the [`dynamo_kv_event_create_removed`] will create remove events
 // these call mus be driving by external c++ threads that are consuming the kv events from the
 // c++ executor api
 
-fn dynemo_create_kv_publisher(
+fn dynamo_create_kv_publisher(
     namespace: String,
     component: String,
     worker_id: i64,
@@ -238,7 +238,7 @@ fn kv_event_create_removed_from_parts(
 /// parent_hash is passed as pointer to indicate whether the blocks
 /// has a parent hash or not. nullptr is used to represent no parent hash
 #[no_mangle]
-pub unsafe extern "C" fn dynemo_kv_event_publish_stored(
+pub unsafe extern "C" fn dynamo_kv_event_publish_stored(
     event_id: u64,
     token_ids: *const u32,
     num_block_tokens: *const usize,
@@ -246,7 +246,7 @@ pub unsafe extern "C" fn dynemo_kv_event_publish_stored(
     num_blocks: usize,
     parent_hash: *const u64,
     lora_id: u64,
-) -> DynemoLlmResult {
+) -> DynamoLlmResult {
     let publisher = KV_PUB.get().unwrap();
     let parent_hash = {
         if parent_hash.is_null() {
@@ -265,40 +265,40 @@ pub unsafe extern "C" fn dynemo_kv_event_publish_stored(
         lora_id,
     );
     match publisher.publish(event) {
-        Ok(_) => DynemoLlmResult::OK,
+        Ok(_) => DynamoLlmResult::OK,
         Err(e) => {
             eprintln!("Error publishing stored kv event {:?}", e);
-            DynemoLlmResult::ERR
+            DynamoLlmResult::ERR
         }
     }
 }
 
 #[no_mangle]
-pub extern "C" fn dynemo_kv_event_publish_removed(
+pub extern "C" fn dynamo_kv_event_publish_removed(
     event_id: u64,
     block_ids: *const u64,
     num_blocks: usize,
-) -> DynemoLlmResult {
+) -> DynamoLlmResult {
     let publisher = KV_PUB.get().unwrap();
     let event = kv_event_create_removed_from_parts(event_id, block_ids, num_blocks);
     match publisher.publish(event) {
-        Ok(_) => DynemoLlmResult::OK,
+        Ok(_) => DynamoLlmResult::OK,
         Err(e) => {
             eprintln!("Error publishing removed kv event {:?}", e);
-            DynemoLlmResult::ERR
+            DynamoLlmResult::ERR
         }
     }
 }
 
 // #[no_mangle]
-// pub extern "C" fn dynemo_kv_publish_store_event(
+// pub extern "C" fn dynamo_kv_publish_store_event(
 //     event_id: u64,
 //     token_ids: *const u32,
 //     num_tokens: usize,
 //     lora_id: u64,
-// ) -> DynemoLlmResult {
+// ) -> DynamoLlmResult {
 //     // if event.is_null() || token_ids.is_null() {
-//     //     return dynemoKvErrorType::INVALID_TOKEN_IDS;
+//     //     return dynamoKvErrorType::INVALID_TOKEN_IDS;
 //     // }
 
 //     // let tokens = unsafe { std::slice::from_raw_parts(token_ids, num_tokens) }.to_vec();
@@ -311,15 +311,15 @@ pub extern "C" fn dynemo_kv_event_publish_removed(
 
 //     // unsafe { *event = Box::into_raw(new_event) };
 
-//     DynemoLlmResult::OK
+//     DynamoLlmResult::OK
 // }
 
 // #[no_mangle]
-// pub extern "C" fn dynemo_kv_event_create_removed(
+// pub extern "C" fn dynamo_kv_event_create_removed(
 //     event_id: u64,
 //     block_hashes: *const u64,
 //     num_hashes: usize,
-// ) -> DynemoLlmResult {
+// ) -> DynamoLlmResult {
 //     // if event.is_null() || block_hashes.is_null() {
 //     //     return -1;
 //     // }
@@ -334,19 +334,19 @@ pub extern "C" fn dynemo_kv_event_publish_removed(
 
 //     // unsafe { *event = Box::into_raw(new_event) };
 //     // 0
-//     DynemoLlmResult::OK
+//     DynamoLlmResult::OK
 // }
 
 // /// create load publisher object and return a handle
 // /// load publisher will instantiate the nats service and tie its stats handler to
 // /// a watch channel receiver.  the watch channel sender will be attach to the
-// /// handle and calls to [`dynemo_load_stats_publish`] issue the stats to the watch t
-// pub extern "C" fn dynemo_load_publisher_create() -> *mut LoadPublisher {
+// /// handle and calls to [`dynamo_load_stats_publish`] issue the stats to the watch t
+// pub extern "C" fn dynamo_load_publisher_create() -> *mut LoadPublisher {
 //     // let publisher = Box::new(LoadPublisher::new());
 //     // Box::into_raw(publisher)
 // }
 
-// pub extern "C" fn dynemo_load_stats_publish(
+// pub extern "C" fn dynamo_load_stats_publish(
 //     publisher: *mut LoadPublisher,
 //     active_slots: u64,
 //     total_slots: u64,

lib/bindings/python/.gitignore

 /target
-python/dynemo/.*.so
+python/dynamo/.*.so

lib/bindings/python/Cargo.lock

@@ -957,7 +957,7 @@ dependencies = [
 ]
 
 [[package]]
-name = "dynemo-llm"
+name = "dynamo-llm"
 version = "0.2.1"
 dependencies = [
  "anyhow",
@@ -972,7 +972,7 @@ dependencies = [
  "chrono",
  "cmake",
  "derive_builder",
- "dynemo-runtime",
+ "dynamo-runtime",
  "either",
  "erased-serde",
  "futures",
@@ -1008,11 +1008,11 @@ dependencies = [
 ]
 
 [[package]]
-name = "dynemo-py3"
+name = "dynamo-py3"
 version = "0.2.1"
 dependencies = [
- "dynemo-llm",
- "dynemo-runtime",
+ "dynamo-llm",
+ "dynamo-runtime",
  "futures",
  "once_cell",
  "pyo3",
@@ -1028,7 +1028,7 @@ dependencies = [
 ]
 
 [[package]]
-name = "dynemo-runtime"
+name = "dynamo-runtime"
 version = "0.2.1"
 dependencies = [
  "anyhow",

lib/bindings/python/Cargo.toml

@@ -14,7 +14,7 @@
 # limitations under the License.
 
 [package]
-name = "dynemo-py3"
+name = "dynamo-py3"
 version = "0.2.1"
 edition = "2021"
 authors = ["NVIDIA"]
@@ -30,8 +30,9 @@ crate-type = ["cdylib"]
 
 
 [dependencies]
-dynemo-llm = { path = "../../llm", features = ["python"] }
-dynemo-runtime = { path = "../../runtime" }
+
+dynamo-llm = { path = "../../llm", features = ["python"] }
+dynamo-runtime = { path = "../../runtime" }
 
 futures = "0.3"
 once_cell = "1.20.3"

lib/bindings/python/README.md

@@ -41,7 +41,7 @@ source .venv/bin/activate
 uv pip install maturin
 ```
 
-4. Build and install dynemo wheel
+4. Build and install dynamo wheel
 ```
 maturin develop --uv
 ```

lib/bindings/python/examples/bls/bar.py

@@ -17,7 +17,7 @@ import asyncio
 
 import uvloop
 
-from dynemo.runtime import DistributedRuntime, dynemo_worker
+from dynamo.runtime import DistributedRuntime, dynamo_worker
 
 uvloop.install()
 
@@ -29,7 +29,7 @@ class RequestHandler:
             yield char
 
 
-@dynemo_worker()
+@dynamo_worker()
 async def worker(runtime: DistributedRuntime):
     component = runtime.namespace("examples/bls").component("bar")
     await component.create_service()

lib/bindings/python/examples/bls/bls.py

@@ -17,12 +17,12 @@ import asyncio
 
 import uvloop
 
-from dynemo.runtime import DistributedRuntime, dynemo_worker
+from dynamo.runtime import DistributedRuntime, dynamo_worker
 
 uvloop.install()
 
 
-@dynemo_worker()
+@dynamo_worker()
 async def worker(runtime: DistributedRuntime):
     foo = (
         await runtime.namespace("examples/bls")