@aws/ml-container-creator 0.2.5 → 0.3.0

package/servers/{model-picker → lib}/catalogs/popular-diffusors.json

@@ -3,7 +3,11 @@
         "family": "stable-diffusion-3",
         "chat_template": null,
         "gated": false,
-        "tags": ["image-generation", "diffusion", "stable-diffusion"],
+        "tags": [
+            "image-generation",
+            "diffusion",
+            "stable-diffusion"
+        ],
         "architecture": "StableDiffusion3Pipeline",
         "framework_compatibility": {
             "vllm-omni": ">=0.14.0"
@@ -12,8 +16,7 @@
         "profiles": {
             "default": {
                 "displayName": "SD3.5 Medium",
-                "envVars": {},
-                "recommendedInstanceTypes": ["ml.g5.2xlarge", "ml.g5.4xlarge"]
+                "envVars": {}
             }
         },
         "notes": "Stable Diffusion 3.5 medium model. Supported natively by vLLM-Omni StableDiffusion3Pipeline."
@@ -22,7 +25,11 @@
         "family": "flux",
         "chat_template": null,
         "gated": true,
-        "tags": ["image-generation", "diffusion", "flux"],
+        "tags": [
+            "image-generation",
+            "diffusion",
+            "flux"
+        ],
         "architecture": "FluxPipeline",
         "framework_compatibility": {
             "vllm-omni": ">=0.14.0"
@@ -31,8 +38,7 @@
         "profiles": {
             "default": {
                 "displayName": "FLUX.1 Dev",
-                "envVars": {},
-                "recommendedInstanceTypes": ["ml.g5.4xlarge", "ml.g5.12xlarge"]
+                "envVars": {}
             }
         },
         "notes": "FLUX.1-dev high-quality generation model. Uses dual text encoders (CLIP + T5) and FlowMatchEuler scheduler. Requires significant VRAM."
@@ -41,7 +47,11 @@
         "family": "flux",
         "chat_template": null,
         "gated": false,
-        "tags": ["image-generation", "diffusion", "flux"],
+        "tags": [
+            "image-generation",
+            "diffusion",
+            "flux"
+        ],
         "architecture": "FluxPipeline",
         "framework_compatibility": {
             "vllm-omni": ">=0.14.0"
@@ -53,7 +63,11 @@
         "family": "wan",
         "chat_template": null,
         "gated": false,
-        "tags": ["video-generation", "diffusion", "wan"],
+        "tags": [
+            "video-generation",
+            "diffusion",
+            "wan"
+        ],
         "architecture": "WanPipeline",
         "framework_compatibility": {
             "vllm-omni": ">=0.16.0"
@@ -65,7 +79,11 @@
         "family": "stable-diffusion",
         "chat_template": null,
         "gated": false,
-        "tags": ["image-generation", "diffusion", "stable-diffusion"],
+        "tags": [
+            "image-generation",
+            "diffusion",
+            "stable-diffusion"
+        ],
         "architecture": null,
         "framework_compatibility": {
             "vllm-omni": ">=0.14.0"
@@ -77,7 +95,11 @@
         "family": "flux",
         "chat_template": null,
         "gated": false,
-        "tags": ["image-generation", "diffusion", "flux"],
+        "tags": [
+            "image-generation",
+            "diffusion",
+            "flux"
+        ],
         "architecture": null,
         "framework_compatibility": {
             "vllm-omni": ">=0.14.0"

package/servers/{model-picker → lib}/catalogs/popular-transformers.json

@@ -3,7 +3,11 @@
         "family": "gpt-oss",
         "chat_template": "",
         "gated": false,
-        "tags": ["text-generation", "openai", "conversational"],
+        "tags": [
+            "text-generation",
+            "openai",
+            "conversational"
+        ],
         "architecture": "GPT2LMHeadModel",
         "framework_compatibility": {
             "vllm": ">=0.3.0",
@@ -17,7 +21,11 @@
         "family": "llama-2",
         "chat_template": "{% for message in messages %}{% if message['role'] == 'system' %}{{ '[INST] <<SYS>>\\n' + message['content'] + '\\n<</SYS>>\\n\\n' }}{% elif message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ ' ' + message['content'] + ' ' }}{% endif %}{% endfor %}",
         "gated": true,
-        "tags": ["text-generation", "llama-2", "conversational"],
+        "tags": [
+            "text-generation",
+            "llama-2",
+            "conversational"
+        ],
         "architecture": "LlamaForCausalLM",
         "framework_compatibility": {
             "vllm": ">=0.3.0",
@@ -31,8 +39,7 @@
                 "envVars": {
                     "MAX_MODEL_LEN": "4096",
                     "GPU_MEMORY_UTILIZATION": "0.9"
-                },
-                "recommendedInstanceTypes": ["ml.g5.xlarge", "ml.g5.2xlarge"]
+                }
             }
         },
         "notes": "Llama-2 7B chat model with official chat template. Requires HuggingFace authentication for download"
@@ -41,7 +48,11 @@
         "family": "llama-2",
         "chat_template": "{% for message in messages %}{% if message['role'] == 'system' %}{{ '[INST] <<SYS>>\\n' + message['content'] + '\\n<</SYS>>\\n\\n' }}{% elif message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ ' ' + message['content'] + ' ' }}{% endif %}{% endfor %}",
         "gated": true,
-        "tags": ["text-generation", "llama-2", "conversational"],
+        "tags": [
+            "text-generation",
+            "llama-2",
+            "conversational"
+        ],
         "architecture": "LlamaForCausalLM",
         "framework_compatibility": {
             "vllm": ">=0.3.0",
@@ -55,18 +66,20 @@
                 "envVars": {
                     "MAX_MODEL_LEN": "4096",
                     "GPU_MEMORY_UTILIZATION": "0.9"
-                },
-                "recommendedInstanceTypes": ["ml.g5.2xlarge", "ml.g5.4xlarge"]
+                }
             }
         },
         "notes": "Llama-2 13B chat model. Requires more GPU memory than 7B variant"
     },
-
     "meta-llama/Llama-2-70b-chat-hf": {
         "family": "llama-2",
         "chat_template": "{% for message in messages %}{% if message['role'] == 'system' %}{{ '[INST] <<SYS>>\\n' + message['content'] + '\\n<</SYS>>\\n\\n' }}{% elif message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ ' ' + message['content'] + ' ' }}{% endif %}{% endfor %}",
         "gated": true,
-        "tags": ["text-generation", "llama-2", "conversational"],
+        "tags": [
+            "text-generation",
+            "llama-2",
+            "conversational"
+        ],
         "architecture": "LlamaForCausalLM",
         "framework_compatibility": {
             "vllm": ">=0.3.0",
@@ -81,8 +94,7 @@
                     "TENSOR_PARALLEL_SIZE": "2",
                     "MAX_MODEL_LEN": "4096",
                     "GPU_MEMORY_UTILIZATION": "0.95"
-                },
-                "recommendedInstanceTypes": ["ml.g5.12xlarge"]
+                }
             },
             "70b-tp4": {
                 "displayName": "Llama-2 70B (4-GPU)",
@@ -90,8 +102,7 @@
                     "TENSOR_PARALLEL_SIZE": "4",
                     "MAX_MODEL_LEN": "4096",
                     "GPU_MEMORY_UTILIZATION": "0.9"
-                },
-                "recommendedInstanceTypes": ["ml.g5.12xlarge", "ml.g5.48xlarge"]
+                }
             }
         },
         "notes": "Llama-2 70B requires tensor parallelism across multiple GPUs"
@@ -100,7 +111,11 @@
         "family": "mistral",
         "chat_template": "{{ bos_token }}{% for message in messages %}{% if message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ message['content'] + eos_token }}{% endif %}{% endfor %}",
         "gated": false,
-        "tags": ["text-generation", "mistral", "conversational"],
+        "tags": [
+            "text-generation",
+            "mistral",
+            "conversational"
+        ],
         "architecture": "MistralForCausalLM",
         "framework_compatibility": {
             "vllm": ">=0.3.0",
@@ -114,8 +129,7 @@
                 "envVars": {
                     "MAX_MODEL_LEN": "8192",
                     "GPU_MEMORY_UTILIZATION": "0.9"
-                },
-                "recommendedInstanceTypes": ["ml.g5.xlarge", "ml.g5.2xlarge"]
+                }
             }
         },
         "notes": "Mistral 7B v0.1 with 8K context window"
@@ -124,7 +138,11 @@
         "family": "mistral",
         "chat_template": "{{ bos_token }}{% for message in messages %}{% if message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ message['content'] + eos_token }}{% endif %}{% endfor %}",
         "gated": false,
-        "tags": ["text-generation", "mistral", "conversational"],
+        "tags": [
+            "text-generation",
+            "mistral",
+            "conversational"
+        ],
         "architecture": "MistralForCausalLM",
         "framework_compatibility": {
             "vllm": ">=0.3.0",
@@ -138,8 +156,7 @@
                 "envVars": {
                     "MAX_MODEL_LEN": "32768",
                     "GPU_MEMORY_UTILIZATION": "0.9"
-                },
-                "recommendedInstanceTypes": ["ml.g5.2xlarge", "ml.g5.4xlarge"]
+                }
             }
         },
         "notes": "Mistral 7B v0.2 with extended 32K context window. Requires more memory for long contexts"
@@ -148,7 +165,11 @@
         "family": "mistral",
         "chat_template": "{{ bos_token }}{% for message in messages %}{% if message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ message['content'] + eos_token }}{% endif %}{% endfor %}",
         "gated": false,
-        "tags": ["text-generation", "mistral", "mixture-of-experts"],
+        "tags": [
+            "text-generation",
+            "mistral",
+            "mixture-of-experts"
+        ],
         "architecture": "MixtralForCausalLM",
         "framework_compatibility": {
             "vllm": ">=0.3.0",
@@ -163,8 +184,7 @@
                     "TENSOR_PARALLEL_SIZE": "2",
                     "MAX_MODEL_LEN": "32768",
                     "GPU_MEMORY_UTILIZATION": "0.95"
-                },
-                "recommendedInstanceTypes": ["ml.g5.12xlarge"]
+                }
             }
         },
         "notes": "Mixtral 8x7B MoE model. Requires tensor parallelism for efficient inference"
@@ -173,7 +193,10 @@
         "family": "llama-2",
         "chat_template": "{% for message in messages %}{% if message['role'] == 'system' %}{{ '[INST] <<SYS>>\\n' + message['content'] + '\\n<</SYS>>\\n\\n' }}{% elif message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ ' ' + message['content'] + ' ' }}{% endif %}{% endfor %}",
         "gated": true,
-        "tags": ["text-generation", "llama-2"],
+        "tags": [
+            "text-generation",
+            "llama-2"
+        ],
         "architecture": null,
         "framework_compatibility": {
             "vllm": ">=0.3.0",
@@ -187,7 +210,10 @@
         "family": "mistral",
         "chat_template": "{{ bos_token }}{% for message in messages %}{% if message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ message['content'] + eos_token }}{% endif %}{% endfor %}",
         "gated": false,
-        "tags": ["text-generation", "mistral"],
+        "tags": [
+            "text-generation",
+            "mistral"
+        ],
         "architecture": null,
         "framework_compatibility": {
             "vllm": ">=0.3.0",
@@ -201,7 +227,11 @@
         "family": "codellama",
         "chat_template": "{% for message in messages %}{% if message['role'] == 'system' %}{{ '[INST] <<SYS>>\\n' + message['content'] + '\\n<</SYS>>\\n\\n' }}{% elif message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ ' ' + message['content'] + ' ' }}{% endif %}{% endfor %}",
         "gated": false,
-        "tags": ["text-generation", "code", "codellama"],
+        "tags": [
+            "text-generation",
+            "code",
+            "codellama"
+        ],
         "architecture": null,
         "framework_compatibility": {
             "vllm": ">=0.3.0",
@@ -214,7 +244,10 @@
         "family": "falcon",
         "chat_template": null,
         "gated": false,
-        "tags": ["text-generation", "falcon"],
+        "tags": [
+            "text-generation",
+            "falcon"
+        ],
         "architecture": null,
         "framework_compatibility": {
             "vllm": ">=0.3.0",

package/servers/{base-image-picker → lib}/catalogs/python-slim.json

@@ -1,38 +1,38 @@
 [
     {
-        "image": "python:3.12-slim",
+        "image": "public.ecr.aws/docker/library/python:3.12-slim",
         "tag": "3.12-slim",
         "architecture": "amd64",
         "created": "2024-10-01T00:00:00Z",
         "labels": { "python_version": "3.12" },
-        "registry": "dockerhub",
-        "repository": "python"
+        "registry": "ecr-public",
+        "repository": "docker/library/python"
     },
     {
-        "image": "python:3.11-slim",
+        "image": "public.ecr.aws/docker/library/python:3.11-slim",
         "tag": "3.11-slim",
         "architecture": "amd64",
         "created": "2023-10-01T00:00:00Z",
         "labels": { "python_version": "3.11" },
-        "registry": "dockerhub",
-        "repository": "python"
+        "registry": "ecr-public",
+        "repository": "docker/library/python"
     },
     {
-        "image": "python:3.10-slim",
+        "image": "public.ecr.aws/docker/library/python:3.10-slim",
         "tag": "3.10-slim",
         "architecture": "amd64",
         "created": "2022-10-01T00:00:00Z",
         "labels": { "python_version": "3.10" },
-        "registry": "dockerhub",
-        "repository": "python"
+        "registry": "ecr-public",
+        "repository": "docker/library/python"
     },
     {
-        "image": "python:3.9-slim",
+        "image": "public.ecr.aws/docker/library/python:3.9-slim",
         "tag": "3.9-slim",
         "architecture": "amd64",
         "created": "2021-10-01T00:00:00Z",
         "labels": { "python_version": "3.9" },
-        "registry": "dockerhub",
-        "repository": "python"
+        "registry": "ecr-public",
+        "repository": "docker/library/python"
     }
 ]

package/servers/lib/schemas/image-catalog.schema.json

@@ -63,12 +63,6 @@
                         },
                         "inferenceAmiVersion": {
                             "type": "string"
-                        },
-                        "recommendedInstanceTypes": {
-                            "type": "array",
-                            "items": {
-                                "type": "string"
-                            }
                         }
                     },
                     "additionalProperties": false
@@ -142,12 +136,6 @@
                                     "type": "string"
                                 }
                             },
-                            "recommendedInstanceTypes": {
-                                "type": "array",
-                                "items": {
-                                    "type": "string"
-                                }
-                            },
                             "notes": {
                                 "type": "string"
                             }
@@ -157,6 +145,12 @@
                 },
                 "notes": {
                     "type": "string"
+                },
+                "supportedModelTypes": {
+                    "type": "array",
+                    "items": {
+                        "type": "string"
+                    }
                 }
             },
             "additionalProperties": false

package/servers/lib/schemas/instances.schema.json

@@ -92,6 +92,35 @@
                     },
                     "notes": {
                         "type": "string"
+                    },
+                    "gpuMemoryGb": {
+                        "oneOf": [
+                            {
+                                "type": "number",
+                                "minimum": 0
+                            },
+                            {
+                                "type": "null"
+                            }
+                        ]
+                    },
+                    "gpuType": {
+                        "oneOf": [
+                            {
+                                "type": "string"
+                            },
+                            {
+                                "type": "null"
+                            }
+                        ]
+                    },
+                    "costTier": {
+                        "type": "string",
+                        "enum": [
+                            "low",
+                            "medium",
+                            "high"
+                        ]
                     }
                 },
                 "additionalProperties": false

package/servers/lib/schemas/model-catalog.schema.json

@@ -21,8 +21,12 @@
             },
             "chat_template": {
                 "oneOf": [
-                    { "type": "string" },
-                    { "type": "null" }
+                    {
+                        "type": "string"
+                    },
+                    {
+                        "type": "null"
+                    }
                 ]
             },
             "gated": {
@@ -36,8 +40,12 @@
             },
             "architecture": {
                 "oneOf": [
-                    { "type": "string" },
-                    { "type": "null" }
+                    {
+                        "type": "string"
+                    },
+                    {
+                        "type": "null"
+                    }
                 ]
             },
             "framework_compatibility": {
@@ -74,12 +82,6 @@
                             "additionalProperties": {
                                 "type": "string"
                             }
-                        },
-                        "recommendedInstanceTypes": {
-                            "type": "array",
-                            "items": {
-                                "type": "string"
-                            }
                         }
                     },
                     "additionalProperties": false

package/servers/lib/schemas/unified-model-catalog.schema.json

@@ -0,0 +1,129 @@
+{
+    "$schema": "http://json-schema.org/draft-07/schema#",
+    "$id": "unified-model-catalog.schema.json",
+    "type": "object",
+    "minProperties": 1,
+    "additionalProperties": {
+        "type": "object",
+        "required": [
+            "architecture",
+            "tasks",
+            "modelType"
+        ],
+        "properties": {
+            "architecture": {
+                "oneOf": [
+                    {
+                        "type": "string",
+                        "minLength": 1
+                    },
+                    {
+                        "type": "null"
+                    }
+                ]
+            },
+            "tasks": {
+                "type": "array",
+                "items": {
+                    "type": "string"
+                },
+                "minItems": 1
+            },
+            "modelType": {
+                "type": "string",
+                "enum": [
+                    "transformer",
+                    "diffusor",
+                    "predictor"
+                ]
+            },
+            "family": {
+                "type": "string"
+            },
+            "parameterCount": {
+                "type": "integer"
+            },
+            "defaultDtype": {
+                "type": "string",
+                "enum": [
+                    "float32",
+                    "float16",
+                    "bfloat16",
+                    "int8",
+                    "int4"
+                ]
+            },
+            "maxPositionEmbeddings": {
+                "type": "integer"
+            },
+            "recommendedQuantizations": {
+                "type": "array",
+                "items": {
+                    "type": "string"
+                }
+            },
+            "chatTemplate": {
+                "oneOf": [
+                    {
+                        "type": "string"
+                    },
+                    {
+                        "type": "null"
+                    }
+                ]
+            },
+            "gated": {
+                "type": "boolean"
+            },
+            "tags": {
+                "type": "array",
+                "items": {
+                    "type": "string"
+                }
+            },
+            "frameworkCompatibility": {
+                "type": "object",
+                "additionalProperties": {
+                    "type": "string"
+                }
+            },
+            "validationLevel": {
+                "type": "string",
+                "enum": [
+                    "tested",
+                    "community-validated",
+                    "experimental",
+                    "untested"
+                ]
+            },
+            "profiles": {
+                "type": "object",
+                "additionalProperties": {
+                    "type": "object",
+                    "required": [
+                        "displayName"
+                    ],
+                    "properties": {
+                        "displayName": {
+                            "type": "string"
+                        },
+                        "envVars": {
+                            "type": "object",
+                            "additionalProperties": {
+                                "type": "string"
+                            }
+                        }
+                    },
+                    "additionalProperties": false
+                }
+            },
+            "notes": {
+                "type": "string"
+            },
+            "pipeline": {
+                "type": "string"
+            }
+        },
+        "additionalProperties": false
+    }
+}

package/servers/model-picker/index.js

@@ -195,11 +195,12 @@ class HuggingFaceResolver extends ModelResolver {
         }
 
         // Fetch model config (conditional)
-        if (!fields || fields.includes('architecture')) {
+        if (!fields || fields.includes('architecture') || fields.includes('model_type')) {
             const modelConfig = await this._fetchJson(
                 `${this.baseUrl}/${modelId}/resolve/main/config.json`
             )
             metadata.architecture = modelConfig?.architectures?.[0] || null
+            metadata.model_type = modelConfig?.model_type || null
         }
 
         return Object.keys(metadata).length > 0 ? metadata : null
@@ -1433,9 +1434,8 @@ let POPULAR_MODELS_CATALOG
 
 try {
     POPULAR_MODELS_CATALOG = {
-        ...loadCatalog('./catalogs/popular-transformers.json'),
-        ...loadCatalog('./catalogs/popular-diffusors.json'),
-        ...loadCatalog('./catalogs/jumpstart-public.json')
+        ...loadCatalog('../lib/catalogs/models.json'),
+        ...loadCatalog('../lib/catalogs/jumpstart-public.json')
     }
 } catch (err) {
     process.stderr.write(`[model-picker] Fatal: ${err.message}\n`)

package/servers/model-picker/manifest.json

@@ -8,9 +8,8 @@
         "discover": true
     },
     "catalogs": {
-        "popular-transformers": "./catalogs/popular-transformers.json",
-        "popular-diffusors": "./catalogs/popular-diffusors.json",
-        "jumpstart-public": "./catalogs/jumpstart-public.json"
+        "models": "../lib/catalogs/models.json",
+        "jumpstart-public": "../lib/catalogs/jumpstart-public.json"
     },
     "tool": {
         "name": "get_models"

package/servers/region-picker/index.js

@@ -64,7 +64,7 @@ let AWS_REGIONS
 let VALID_REGION_CODES
 
 try {
-    AWS_REGIONS = loadCatalog('./catalogs/regions.json')
+    AWS_REGIONS = loadCatalog('../lib/catalogs/regions.json')
     VALID_REGION_CODES = new Set(AWS_REGIONS.map(r => r.code))
 } catch (err) {
     process.stderr.write(`[region-picker] Fatal: ${err.message}\n`)

package/servers/region-picker/manifest.json

@@ -8,7 +8,7 @@
         "discover": false
     },
     "catalogs": {
-        "regions": "./catalogs/regions.json"
+        "regions": "../lib/catalogs/regions.json"
     },
     "tool": {
         "name": "get_regions"