diff --git a/third_party/Dell/model-deployment/README.md b/third_party/Dell/model-deployment/README.md deleted file mode 100644 index 43d98118..00000000 --- a/third_party/Dell/model-deployment/README.md +++ /dev/null @@ -1 +0,0 @@ -# PLACEHOLDER \ No newline at end of file diff --git a/third_party/Dell/model-deployment/TinyLlama-1.1B-Chat-v1.0/deployment.md b/third_party/Dell/model-deployment/TinyLlama-1.1B-Chat-v1.0/deployment.md new file mode 100644 index 00000000..4ed9503b --- /dev/null +++ b/third_party/Dell/model-deployment/TinyLlama-1.1B-Chat-v1.0/deployment.md @@ -0,0 +1,113 @@ +## Step 1: Prerequisites to Deploy TinyLlama Model on Xeon with Keycloak + +Ensure the Enterprise Inference stack with Keycloak is already deployed before proceeding. + +Edit `core/scripts/generate-token.sh` and set your values before sourcing it: + +| Variable | Description | +| ------------------------- | ------------------------------------------------------------------------ | +| `BASE_URL` | Hostname of your cluster (e.g. `api.example.com`), without `https://` | +| `KEYCLOAK_ADMIN_USERNAME` | Keycloak admin username | +| `KEYCLOAK_PASSWORD` | Keycloak admin password | +| `KEYCLOAK_CLIENT_ID` | Keycloak client ID configured during EI deployment | + +Then run: + +```bash +export HUGGING_FACE_HUB_TOKEN="your_token_here" + +cd ~/Enterprise-Inference +source core/scripts/generate-token.sh +``` + +This exports: `BASE_URL`, `KEYCLOAK_CLIENT_ID`, `KEYCLOAK_CLIENT_SECRET`, and `TOKEN`. + +## Step 2: Deploy Tinyllama-1.1b-chat-v1.0 Model + +```bash +helm install tinyllama-1-1b-cpu ./core/helm-charts/vllm \ + --values ./core/helm-charts/vllm/xeon-values.yaml \ + --set LLM_MODEL_ID="TinyLlama/TinyLlama-1.1B-Chat-v1.0" \ + --set global.HUGGINGFACEHUB_API_TOKEN="$HUGGING_FACE_HUB_TOKEN" \ + --set ingress.enabled=true \ + --set ingress.secretname="${BASE_URL}" \ + --set ingress.host="${BASE_URL}" \ + --set oidc.client_id="$KEYCLOAK_CLIENT_ID" \ + --set oidc.client_secret="$KEYCLOAK_CLIENT_SECRET" \ + --set apisix.enabled=true \ + --set tensor_parallel_size="1" \ + --set pipeline_parallel_size="1" +``` + +## Step 3: Verify the Deployment + +```bash +kubectl get pods +kubectl get apisixroutes +``` + +Expected Output: + +``` +NAME READY STATUS RESTARTS +keycloak-0 1/1 Running 0 +keycloak-postgresql-0 1/1 Running 0 +tinyllama-1-1b-cpu-vllm-- 1/1 Running 0 +``` + +> Note: The pod name suffix `-` is auto-generated by Kubernetes and will differ on each deployment. Ensure all pods show `1/1 Running`. + +``` +NAME HOSTS +tinyllama-1-1b-cpu-vllm-apisixroute api.example.com +``` + +## Step 4: Test the Deployed Model + +```bash +curl -k https://${BASE_URL}/tinyLlama-1.1B-Chat-v1.0-vllmcpu/v1/completions \ + -X POST \ + -H "Content-Type: application/json" \ + -H "Authorization: Bearer $TOKEN" \ + -d '{ + "model": "TinyLlama/TinyLlama-1.1B-Chat-v1.0", + "prompt": "What is Deep Learning?", + "max_tokens": 25, + "temperature": 0 + }' +``` + +If successful, the model will return a completion response. + +## To undeploy the model + +```bash +helm uninstall tinyllama-1-1b-cpu +``` + +## Parameters + +| Parameter | Description | +| --------------------------------------------------------- | ----------------------------------------------------------------------------------------------------- | +| `--set LLM_MODEL_ID="TinyLlama/TinyLlama-1.1B-Chat-v1.0"` | Defines the target model from **Hugging Face** to deploy. | +| `--set global.HUGGINGFACEHUB_API_TOKEN="..."` | Authenticates access to gated or private Hugging Face models. Replace with your own secure token. | +| `--set ingress.enabled=true` | Enables Kubernetes **Ingress** to expose the model service externally. | +| `--set ingress.host="replace-ingress"` | Public hostname or FQDN for the inference endpoint (maps to your Ingress controller IP). | +| `--set ingress.secretname="replace-secret"` | Kubernetes **TLS Secret** used for HTTPS termination at the ingress layer. | +| `--set oidc.client_id="..."` | Keycloak OIDC client ID used for token-based authentication. | +| `--set oidc.client_secret="..."` | Keycloak OIDC client secret corresponding to the client ID. | +| `--set apisix.enabled=true` | Enables **APISIX** as the API gateway for routing and authentication. | +| `--set tensor_parallel_size="1"` | Number of tensor parallel workers. Set to the number of available CPUs/GPUs per node. | +| `--set pipeline_parallel_size="1"` | Number of pipeline parallel stages. Typically `1` for single-node deployments. | + + + + + + + + + + + + diff --git a/third_party/Dell/model-deployment/TinyLlama-1.1B-Chat-v1.0/model-card.md b/third_party/Dell/model-deployment/TinyLlama-1.1B-Chat-v1.0/model-card.md new file mode 100644 index 00000000..c14ec2e2 --- /dev/null +++ b/third_party/Dell/model-deployment/TinyLlama-1.1B-Chat-v1.0/model-card.md @@ -0,0 +1,60 @@ +# TinyLlama-1.1B-Chat-v1.0 + +This model uses TinyLlama-1.1B-Chat-v1.0, a compact large language model developed by the TinyLlama Project team. It is a chat-tuned variant of the TinyLlama 1.1B base model, optimized for instruction-following, conversational AI, and lightweight reasoning tasks. Despite its small size, TinyLlama delivers strong performance for edge AI, embedded systems, rapid prototyping, and cost-efficient inference scenarios. + +For full details including model specifications, licensing, intended use, safety guidance, and example prompts, please visit the official Hugging Face page: **Official Hugging Face Page** + +https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0 + +This model provides inference services only; weights are hosted by Hugging Face under the Apache 2.0 License. + +Ensure compliance with the Apache 2.0 License terms before using this model. + +### Model Attribution + +**Developer:** TinyLlama Project + +**purpose:** Lightweight instruction-tuned conversational AI + +**Sizes/Variants:** 1.1B parameters + +**Modalities:** Text → Natural Language + +**Parameter Size:** 1.1 Billion + +**Max Context:** ~2K tokens + +**License:** Apache 2.0 (commercial-friendly) + +### Usage Notice + +**By using this model, you agree that:** + +- Inputs and outputs are processed by the TinyLlama-1.1B-Chat-v1.0 model under the Apache 2.0 license. +- You are responsible for validating outputs before production use. +- This model should not be used for generating malicious, deceptive, or unsafe content. +- Outputs may contain inaccuracies and must be reviewed for correctness and compliance. + +### Intended Applications + +- Lightweight chatbots and virtual assistants +- Edge AI and on-device inference +- Rapid prototyping and AI experimentation +- CPU-based conversational agents +- Educational tools and demos +- RAG-based document assistants for low-resource environments +- Dev/test automation helpers + +### Limitations + +- Limited reasoning depth compared to large models (7B+) +- Reduced long-context understanding +- Not suitable for complex multi-step logic or heavy code generation +- May hallucinate or oversimplify responses +- Not designed for safety-critical or regulated decision systems + +### References + +TinyLlama Project — https://github.com/jzhang38/TinyLlama + +Hugging Face Model Page — https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0