Squashed 'ai-context/trustgraph-templates/' content from commit 42a5fd1b

git-subtree-dir: ai-context/trustgraph-templates
git-subtree-split: 42a5fd1b678f32be378062e30451e2052ccb95dd

trustgraph_configurator/resources/dialog/docs/model/azure-compose.md

@@ -0,0 +1,6 @@
+To use Azure Serverless APIs, you need to have a serverless endpoint deployed, and you must also provide an endpoint token as an environment variable.
+
+```
+AZURE_ENDPOINT=https://ENDPOINT.API.HOST.GOES.HERE/
+AZURE_TOKEN=TOKEN-GOES-HERE
+```

trustgraph_configurator/resources/dialog/docs/model/azure-k8s.md

@@ -0,0 +1,8 @@
+To use Azure Serverless APIs, you need to have a serverless endpoint deployed. You must also provide an Azure endpoint and token in a Kubernetes secret before launching the application.
+
+```bash
+kubectl -n {{namespace}} create secret \
+    generic azure-credentials \
+    --from-literal=azure-endpoint=AZURE-ENDPOINT \
+    --from-literal=azure-token=AZURE-TOKEN
+```

trustgraph_configurator/resources/dialog/docs/model/azure-openai-compose.md

@@ -0,0 +1,8 @@
+To use Azure's OpenAI APIs, you need to have a serverless OpenAI endpoint deployed, and you must also provide an endpoint token as an environment variable. In addition, the OpenAI API requires an API Version and Model Name to be set. The Model Name is set by the user during the deployment within AzureAI.
+
+```
+AZURE_ENDPOINT=https://ENDPOINT.API.HOST.GOES.HERE/
+AZURE_TOKEN=TOKEN-GOES-HERE
+AZURE_API_VERSION=API-VERSION-GOES-HERE
+AZURE_MODEL=MODEL-NAME-GOES-HERE
+```

trustgraph_configurator/resources/dialog/docs/model/azure-openai-k8s.md

@@ -0,0 +1,10 @@
+To use Azure's OpenAI APIs, you need to have a serverless OpenAI endpoint deployed. You must also provide an endpoint token, API version, and model name in a Kubernetes secret. The Model Name is set by the user during the deployment within AzureAI.
+
+```bash
+kubectl -n {{namespace}} create secret \
+    generic azure-openai-credentials \
+    --from-literal=azure-endpoint=https://ENDPOINT.API.HOST.GOES.HERE/ \
+    --from-literal=azure-token=TOKEN-GOES-HERE \
+    --from-literal=azure-api-version=API-VERSION-GOES-HERE \
+    --from-literal=azure-model=MODEL-NAME-GOES-HERE
+```

trustgraph_configurator/resources/dialog/docs/model/bedrock-compose.md

@@ -0,0 +1,7 @@
+To use AWS Bedrock, you must have enabled models in the AWS Bedrock console. You must also provide an AWS access key ID and secret key.
+
+```
+AWS_ACCESS_KEY_ID=ID-KEY-HERE
+AWS_SECRET_ACCESS_KEY=TOKEN-GOES-HERE
+AWS_DEFAULT_REGION=AWS-REGION-HERE
+```

trustgraph_configurator/resources/dialog/docs/model/bedrock-k8s.md

@@ -0,0 +1,9 @@
+To use AWS Bedrock, you must have enabled models in the AWS Bedrock console. You must also provide an AWS access key ID and secret key as a Kubernetes secret before deploying the application.
+
+```bash
+kubectl -n {{namespace}} create secret \
+    generic bedrock-credentials \
+    --from-literal=aws-id-key=AWS-ID-KEY \
+    --from-literal=aws-secret=AWS-SECRET-KEY \
+    --from-literal=aws-region=AWS-REGION-HERE
+```

trustgraph_configurator/resources/dialog/docs/model/claude-compose.md

@@ -0,0 +1,5 @@
+To use Anthropic Claude, you need a Claude API key. Provide the Claude API key in an environment variable when running the Docker Compose configuration.
+
+```
+CLAUDE_KEY=CLAUDE-KEY-GOES-HERE
+```

trustgraph_configurator/resources/dialog/docs/model/claude-k8s.md

@@ -0,0 +1,7 @@
+To use Anthropic Claude, you need a Claude API key which must be provided in a Kubernetes secret.
+
+```bash
+kubectl -n {{namespace}} create secret \
+    generic claude-credentials \
+    --from-literal=claude-key=CLAUDE-KEY-HERE
+```

trustgraph_configurator/resources/dialog/docs/model/cohere-compose.md

@@ -0,0 +1,5 @@
+To use Cohere APIs, you need an API token which must be provided in an environment variable.
+
+```
+COHERE_KEY=TOKEN-GOES-HERE
+```

trustgraph_configurator/resources/dialog/docs/model/cohere-k8s.md

@@ -0,0 +1,7 @@
+To use Cohere APIs, you need an API token which must be provided in a Kubernetes secret.
+
+```bash
+kubectl -n {{namespace}} create secret \
+    generic cohere-credentials \
+    --from-literal=cohere-key=COHERE-KEY
+```

trustgraph_configurator/resources/dialog/docs/model/google-ai-studio-compose.md

@@ -0,0 +1,5 @@
+To use Google AI Studio APIs, you need an API token which must be provided in an environment variable.
+
+```
+GOOGLE_AI_STUDIO_KEY=TOKEN-GOES-HERE
+```

trustgraph_configurator/resources/dialog/docs/model/google-ai-studio-k8s.md

@@ -0,0 +1,7 @@
+To use Google AI Studio APIs, you need an API token which must be provided in a Kubernetes secret.
+
+```bash
+kubectl -n {{namespace}} create secret \
+    generic googleaistudio-credentials \
+    --from-literal=google-ai-studio-key=GOOGLEAISTUDIO-KEY
+```

trustgraph_configurator/resources/dialog/docs/model/llamafile-compose.md

@@ -0,0 +1,5 @@
+To use Llamafile, you must have a Llamafile service running on an accessible host. The Llamafile host must be provided in an environment variable.
+
+```
+LLAMAFILE_URL=LLAMAFILE-URL
+```

trustgraph_configurator/resources/dialog/docs/model/llamafile-k8s.md

@@ -0,0 +1,7 @@
+To use Llamafile, you must have a Llamafile service running on an accessible host. The Llamafile host must be provided in a Kubernetes secret.
+
+```bash
+kubectl -n {{namespace}} create secret \
+    generic llamafile-credentials \
+    --from-literal=llamafile-url=http://llamafile:1234/
+```

trustgraph_configurator/resources/dialog/docs/model/lm-studio-compose.md

@@ -0,0 +1,9 @@
+LMStudio allows you to run models locally, with a nice UX. The LMStudio application or service must be running, and have the REST API enabled, and model made available by pulling from the model repository.
+
+Note that LMStudio is a commercial product - a licence is needed for non-personal usage. See [lmstudio.ai/work](https://lmstudio.ai/work).
+
+```
+LMSTUDIO_URL=http://localhost:1234
+```
+
+Replace the URL with the URL of your LMStudio API service.

trustgraph_configurator/resources/dialog/docs/model/lm-studio-k8s.md

@@ -0,0 +1,9 @@
+LMStudio service URL must be provided in a Kubernetes secret.
+
+```bash
+kubectl -n {{namespace}} \
+    create secret generic lmstudio-credentials \
+    --from-literal=lmstudio-url=http://lmstudio:11434/
+```
+
+Replace the URL with the URL of your LMStudio service.

trustgraph_configurator/resources/dialog/docs/model/mistral-compose.md

@@ -0,0 +1,5 @@
+To use Mistral, you need a Mistral API key. Provide the Mistral API key in an environment variable when running the Docker Compose configuration.
+
+```
+MISTRAL_TOKEN=TOKEN-GOES-HERE
+```

trustgraph_configurator/resources/dialog/docs/model/mistral-k8s.md

@@ -0,0 +1,7 @@
+To use Mistral, you need a Mistral API key which must be provided in a Kubernetes secret.
+
+```bash
+kubectl -n {{namespace}} create secret \
+    generic mistral-credentials \
+    --from-literal=mistral-key=MISTRAL-TOKEN
+```

trustgraph_configurator/resources/dialog/docs/model/ollama-compose.md

@@ -0,0 +1,9 @@
+The power of Ollama is the flexibility it provides in Language Model deployments. Being able to run LMs with Ollama enables fully secure AI TrustGraph pipelines that aren't relying on any external APIs. No data is leaving the host environment or network.
+
+The Ollama service must be running, and have required models available using `ollama pull`. The Ollama service URL must be provided in an environment variable.
+
+```
+OLLAMA_HOST=http://ollama-host:11434
+```
+
+Replace the URL with the URL of your Ollama service.

trustgraph_configurator/resources/dialog/docs/model/ollama-k8s.md

@@ -0,0 +1,9 @@
+The Ollama service URL must be provided in a Kubernetes secret.
+
+```bash
+kubectl -n {{namespace}} \
+    create secret generic ollama-credentials \
+    --from-literal=ollama-host=http://ollama:11434/
+```
+
+Replace the URL with the URL of your Ollama service.

trustgraph_configurator/resources/dialog/docs/model/openai-compose.md

@@ -0,0 +1,5 @@
+To use OpenAI APIs, you need an API token which must be provided in an environment variable.
+
+```
+OPENAI_TOKEN=TOKEN-GOES-HERE
+```

trustgraph_configurator/resources/dialog/docs/model/openai-k8s.md

@@ -0,0 +1,7 @@
+To use OpenAI APIs, you need an API token which must be provided in a Kubernetes secret.
+
+```bash
+kubectl -n {{namespace}} create secret \
+    generic openai-credentials \
+    --from-literal=openai-token=OPENAI-TOKEN-HERE
+```

trustgraph_configurator/resources/dialog/docs/model/tgi-compose.md

@@ -0,0 +1,9 @@
+Text Generation Inference (TGI) is Hugging Face's production-ready inference server for LLMs. It provides high-performance text generation with features like continuous batching, tensor parallelism, and optimized attention mechanisms.
+
+The TGI service must be running with the required model loaded. The TGI service URL must be provided in an environment variable.
+
+```
+TGI_BASE_URL=http://tgi-host:8080/v1
+```
+
+Replace the URL with the URL of your TGI service, noting the `v1` suffix for OpenAI-compatible API.

trustgraph_configurator/resources/dialog/docs/model/tgi-k8s.md

@@ -0,0 +1,9 @@
+The TGI service URL must be provided in a Kubernetes secret.
+
+```bash
+kubectl -n {{namespace}} \
+    create secret generic tgi-credentials \
+    --from-literal=tgi-url=http://tgi:8080/v1
+```
+
+Replace the URL with the URL of your TGI service.

trustgraph_configurator/resources/dialog/docs/model/vertex-ai-compose.md

@@ -0,0 +1,7 @@
+To use VertexAI, you need to have a Google Cloud credential file provisioned for a service account which has access to the VertexAI services. This means signing up to GCP and using an existing, or launching a new GCP project. The GCP credential will be a JSON file which should be stored in `vertexai/private.json`.
+
+The credential file is mounted as a volume in Docker Compose, which can cause issues with SELinux if you are running on Linux. Make sure that Docker has access to volume files if this affects you.
+
+```bash
+chcon -Rt svirt_sandbox_file_t vertexai/
+```

trustgraph_configurator/resources/dialog/docs/model/vertex-ai-k8s.md

@@ -0,0 +1,10 @@
+To use VertexAI, you need to have a Google Cloud credential file provisioned for a service account which has access to the VertexAI services. This means signing up to GCP and using an existing, or launching a new GCP project. The GCP credential will be a JSON file which would arrive in a file called `private.json`.
+
+The private.json file should be loaded into Kubernetes as a secret.
+
+```bash
+kubectl -n {{namespace}} create secret \
+    generic vertexai-creds --from-file=private.json=private.json
+```
+
+> **Warning:** Google Cloud private.json files are secrets which potentially provide access to all of your Google Cloud resources. Take great care to ensure that the permissions of the account are minimal, ideally scoped to just AI services.

trustgraph_configurator/resources/dialog/docs/model/vllm-compose.md

@@ -0,0 +1,9 @@
+vLLM is a high-throughput, memory-efficient inference and serving engine for LLMs. Using PagedAttention and continuous batching, vLLM enables fully secure AI TrustGraph pipelines that aren't relying on any external APIs. No data is leaving the host environment or network.
+
+The vLLM service must be running with the required model loaded using `vllm serve`. The vLLM service URL must be provided in an environment variable.
+
+```
+VLLM_BASE_URL=http://vllm-host:8000/v1
+```
+
+Replace the URL with the URL of your vLLM service, noting the `v1` suffix.

trustgraph_configurator/resources/dialog/docs/model/vllm-k8s.md

@@ -0,0 +1,9 @@
+The vLLM service URL must be provided in a Kubernetes secret.
+
+```bash
+kubectl -n {{namespace}} \
+    create secret generic vllm-credentials \
+    --from-literal=vllm-url=http://vllm:8000/v1
+```
+
+Replace the URL with the URL of your vLLM service.