diff --git a/src/sagemaker/image_uri_config/huggingface-llm-neuronx.json b/src/sagemaker/image_uri_config/huggingface-llm-neuronx.json index c86a8dcf98..7b96b60ff8 100644 --- a/src/sagemaker/image_uri_config/huggingface-llm-neuronx.json +++ b/src/sagemaker/image_uri_config/huggingface-llm-neuronx.json @@ -4,7 +4,7 @@ "inf2" ], "version_aliases": { - "0.0": "0.0.24" + "0.0": "0.0.25" }, "versions": { "0.0.16": { @@ -301,6 +301,39 @@ "container_version": { "inf2": "ubuntu22.04" } + }, + "0.0.25": { + "py_versions": [ + "py310" + ], + "registries": { + "ap-northeast-1": "763104351884", + "ap-south-1": "763104351884", + "ap-south-2": "772153158452", + "ap-southeast-1": "763104351884", + "ap-southeast-2": "763104351884", + "ap-southeast-4": "457447274322", + "cn-north-1": "727897471807", + "cn-northwest-1": "727897471807", + "eu-central-1": "763104351884", + "eu-central-2": "380420809688", + "eu-south-2": "503227376785", + "eu-west-1": "763104351884", + "eu-west-3": "763104351884", + "il-central-1": "780543022126", + "sa-east-1": "763104351884", + "us-east-1": "763104351884", + "us-east-2": "763104351884", + "us-gov-east-1": "446045086412", + "us-gov-west-1": "442386744353", + "us-west-2": "763104351884", + "ca-west-1": "204538143572" + }, + "tag_prefix": "2.1.2-optimum0.0.25", + "repository": "huggingface-pytorch-tgi-inference", + "container_version": { + "inf2": "ubuntu22.04" + } } } } diff --git a/src/sagemaker/image_uri_config/huggingface-llm.json b/src/sagemaker/image_uri_config/huggingface-llm.json index b3988362fd..24cbd5ca96 100644 --- a/src/sagemaker/image_uri_config/huggingface-llm.json +++ b/src/sagemaker/image_uri_config/huggingface-llm.json @@ -12,7 +12,7 @@ "1.2": "1.2.0", "1.3": "1.3.3", "1.4": "1.4.5", - "2.0": "2.2.0" + "2.0": "2.3.1" }, "versions": { "0.6.0": { @@ -719,6 +719,53 @@ "container_version": { "gpu": "cu121-ubuntu22.04-v2.0" } + }, + "2.3.1": { + "py_versions": [ + "py311" + ], + "registries": { + "af-south-1": "626614931356", + "il-central-1": "780543022126", + "ap-east-1": "871362719292", + "ap-northeast-1": "763104351884", + "ap-northeast-2": "763104351884", + "ap-northeast-3": "364406365360", + "ap-south-1": "763104351884", + "ap-south-2": "772153158452", + "ap-southeast-1": "763104351884", + "ap-southeast-2": "763104351884", + "ap-southeast-3": "907027046896", + "ap-southeast-4": "457447274322", + "ca-central-1": "763104351884", + "cn-north-1": "727897471807", + "cn-northwest-1": "727897471807", + "eu-central-1": "763104351884", + "eu-central-2": "380420809688", + "eu-north-1": "763104351884", + "eu-west-1": "763104351884", + "eu-west-2": "763104351884", + "eu-west-3": "763104351884", + "eu-south-1": "692866216735", + "eu-south-2": "503227376785", + "me-south-1": "217643126080", + "me-central-1": "914824155844", + "sa-east-1": "763104351884", + "us-east-1": "763104351884", + "us-east-2": "763104351884", + "us-gov-east-1": "446045086412", + "us-gov-west-1": "442386744353", + "us-iso-east-1": "886529160074", + "us-isob-east-1": "094389454867", + "us-west-1": "763104351884", + "us-west-2": "763104351884", + "ca-west-1": "204538143572" + }, + "tag_prefix": "2.4.0-tgi2.3.1", + "repository": "huggingface-pytorch-tgi-inference", + "container_version": { + "gpu": "cu124-ubuntu22.04" + } } } } diff --git a/tests/unit/sagemaker/image_uris/test_huggingface_llm.py b/tests/unit/sagemaker/image_uris/test_huggingface_llm.py index c3fab15936..28525a390c 100644 --- a/tests/unit/sagemaker/image_uris/test_huggingface_llm.py +++ b/tests/unit/sagemaker/image_uris/test_huggingface_llm.py @@ -45,6 +45,7 @@ "2.0.1": "2.1.1-tgi2.0.1-gpu-py310-cu121-ubuntu22.04", "2.0.2": "2.3.0-tgi2.0.2-gpu-py310-cu121-ubuntu22.04", "2.2.0": "2.3.0-tgi2.2.0-gpu-py310-cu121-ubuntu22.04-v2.0", + "2.3.1": "2.4.0-tgi2.3.1-gpu-py311-cu124-ubuntu22.04", }, "inf2": { "0.0.16": "1.13.1-optimum0.0.16-neuronx-py310-ubuntu22.04", @@ -56,6 +57,7 @@ "0.0.22": "2.1.2-optimum0.0.22-neuronx-py310-ubuntu22.04", "0.0.23": "2.1.2-optimum0.0.23-neuronx-py310-ubuntu22.04", "0.0.24": "2.1.2-optimum0.0.24-neuronx-py310-ubuntu22.04", + "0.0.25": "2.1.2-optimum0.0.25-neuronx-py310-ubuntu22.04", }, }