chore: Update README.md

Signed-off-by: Neelay Shah <neelays@nvidia.com>

Author: nnshah1

README.md

@@ -101,7 +101,7 @@ APIs are subject to change.
 
 ### Hello World
 
-[Hello World](./runtime/rust/python-wheel/examples/hello_world)
+[Hello World](./lib/bindings/python/examples/hello_world)
 
 A basic example demonstrating the rust based runtime and python
 bindings.

examples/python_rs/llm/vllm/README.md

@@ -23,9 +23,9 @@ This example demonstrates how to use Triton Distributed to serve large language
 
 Start required services (etcd and NATS):
 
-   Option A: Using [Docker Compose](/runtime/rust/docker-compose.yml) (Recommended)
+   Option A: Using [Docker Compose](/deploy/docker-compose.yml) (Recommended)
    ```bash
-   docker-compose up -d
+   docker compose -f ./deploy/docker-compose.yml up -d
    ```
 
    Option B: Manual Setup
@@ -53,7 +53,7 @@ The example is designed to run in a containerized environment using Triton Distr
 
 ## Deployment
 
-#### 1. HTTP Server
+### 1. HTTP Server
 
 Run the server logging (with debug level logging):
 ```bash
@@ -66,26 +66,58 @@ Add model to the server:
 llmctl http add chat-models deepseek-ai/DeepSeek-R1-Distill-Llama-8B triton-init.vllm.generate
 ```
 
+##### Example Output
+```
++------------+------------------------------------------+-----------+-----------+----------+
+| MODEL TYPE | MODEL NAME                               | NAMESPACE | COMPONENT | ENDPOINT |
++------------+------------------------------------------+-----------+-----------+----------+
+| chat       | deepseek-ai/DeepSeek-R1-Distill-Llama-8B | triton-init | vllm      | generate |
++------------+------------------------------------------+-----------+-----------+----------+
+```
+
 ### 2. Workers
 
 #### 2.1. Monolithic Deployment
 
 In a separate terminal run the vllm worker:
 
 ```bash
+# Activate virtual environment
+source /opt/triton/venv/bin/activate
+
 # Launch worker
 cd /workspace/examples/python_rs/llm/vllm
 python3 -m monolith.worker \
     --model deepseek-ai/DeepSeek-R1-Distill-Llama-8B \
     --enforce-eager
 ```
 
+##### Example Output
+
+```
+INFO 03-02 05:30:36 __init__.py:190] Automatically detected platform cuda.
+WARNING 03-02 05:30:36 nixl.py:43] NIXL is not available
+
+INFO 03-02 05:30:43 config.py:542] This model supports multiple tasks: {'embed', 'score', 'generate', 'classify', 'reward'}. Defaulting to 'generate'.
+INFO 03-02 05:30:43 base_engine.py:43] Initializing engine client
+INFO 03-02 05:30:43 api_server.py:206] Started engine process with PID 1151
+INFO 03-02 05:30:44 config.py:542] This model supports multiple tasks: {'embed', 'score', 'generate', 'classify', 'reward'}. Defaulting to 'generate'.
+
+<SNIP>
+
+INFO 03-02 05:32:20 llm_engine.py:476] init engine (profile, create kv cache, warmup model) took 4.22 seconds
+
+```
+
 #### 2.2. Disaggregated Deployment
 
 This deployment option splits the model serving across prefill and decode workers, enabling more efficient resource utilization.
 
 **Terminal 1 - Prefill Worker:**
 ```bash
+# Activate virtual environment
+source /opt/triton/venv/bin/activate
+
 # Launch prefill worker
 cd /workspace/examples/python_rs/llm/vllm
 VLLM_WORKER_MULTIPROC_METHOD=spawn CUDA_VISIBLE_DEVICES=0 python3 -m disaggregated.prefill_worker \
@@ -97,8 +129,21 @@ VLLM_WORKER_MULTIPROC_METHOD=spawn CUDA_VISIBLE_DEVICES=0 python3 -m disaggregat
     '{"kv_connector":"TritonNcclConnector","kv_role":"kv_producer","kv_rank":0,"kv_parallel_size":2}'
 ```
 
+##### Example Output
+
+```
+INFO 03-02 05:59:44 worker.py:269] the current vLLM instance can use total_gpu_memory (47.54GiB) x gpu_memory_utilization (0.40) = 19.01GiB
+INFO 03-02 05:59:44 worker.py:269] model weights take 14.99GiB; non_torch_memory takes 0.06GiB; PyTorch activation peak memory takes 1.19GiB; the rest of the memory reserved for KV Cache is 2.78GiB.
+INFO 03-02 05:59:44 executor_base.py:110] # CUDA blocks: 1423, # CPU blocks: 2048
+INFO 03-02 05:59:44 executor_base.py:115] Maximum concurrency for 10 tokens per request: 2276.80x
+INFO 03-02 05:59:47 llm_engine.py:476] init engine (profile, create kv cache, warmup model) took 3.41 seconds
+```
+
 **Terminal 2 - Decode Worker:**
 ```bash
+# Activate virtual environment
+source /opt/triton/venv/bin/activate
+
 # Launch decode worker
 cd /workspace/examples/python_rs/llm/vllm
 VLLM_WORKER_MULTIPROC_METHOD=spawn CUDA_VISIBLE_DEVICES=1,2 python3 -m disaggregated.decode_worker \
@@ -112,6 +157,16 @@ VLLM_WORKER_MULTIPROC_METHOD=spawn CUDA_VISIBLE_DEVICES=1,2 python3 -m disaggreg
 
 The disaggregated deployment utilizes separate GPUs for prefill and decode operations, allowing for optimized resource allocation and improved performance. For more details on the disaggregated deployment, please refer to the [vLLM documentation](https://docs.vllm.ai/en/latest/features/disagg_prefill.html).
 
+##### Example Output
+
+```
+INFO 03-02 05:59:44 worker.py:269] the current vLLM instance can use total_gpu_memory (47.54GiB) x gpu_memory_utilization (0.40) = 19.01GiB
+INFO 03-02 05:59:44 worker.py:269] model weights take 14.99GiB; non_torch_memory takes 0.06GiB; PyTorch activation peak memory takes 1.19GiB; the rest of the memory reserved for KV Cache is 2.78GiB.
+INFO 03-02 05:59:44 executor_base.py:110] # CUDA blocks: 1423, # CPU blocks: 2048
+INFO 03-02 05:59:44 executor_base.py:115] Maximum concurrency for 10 tokens per request: 2276.80x
+INFO 03-02 05:59:47 llm_engine.py:476] init engine (profile, create kv cache, warmup model) took 3.41 seconds
+```
+
 
 ### 3. Client
 
@@ -126,7 +181,8 @@ curl localhost:8080/v1/chat/completions \
   }'
 ```
 
-Expected output:
+##### Example Output
+
 ```json
 {
     "id": "5b04e7b0-0dcd-4c45-baa0-1d03d924010c",
@@ -264,7 +320,8 @@ curl localhost:8080/v1/chat/completions   -H "Content-Type: application/json"
     "max_tokens": 30
   }'
 ```
-Expected output:
+##### Example Output
+
 ```json
 {
     "id": "f435d1aa-d423-40a0-a616-00bc428a3e32",