llama-stack 0.0.42__py3-none-any.whl → 0.3.4__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- llama_stack/__init__.py +5 -0
- llama_stack/apis/agents/__init__.py +1 -1
- llama_stack/apis/agents/agents.py +700 -281
- llama_stack/apis/agents/openai_responses.py +1311 -0
- llama_stack/{providers/adapters/memory/sample/config.py → apis/batches/__init__.py} +2 -5
- llama_stack/apis/batches/batches.py +100 -0
- llama_stack/apis/benchmarks/__init__.py +7 -0
- llama_stack/apis/benchmarks/benchmarks.py +108 -0
- llama_stack/apis/common/content_types.py +143 -0
- llama_stack/apis/common/errors.py +103 -0
- llama_stack/apis/common/job_types.py +38 -0
- llama_stack/apis/common/responses.py +36 -0
- llama_stack/apis/common/training_types.py +36 -5
- llama_stack/apis/common/type_system.py +158 -0
- llama_stack/apis/conversations/__init__.py +31 -0
- llama_stack/apis/conversations/conversations.py +286 -0
- llama_stack/apis/datasetio/__init__.py +7 -0
- llama_stack/apis/datasetio/datasetio.py +59 -0
- llama_stack/apis/datasets/__init__.py +7 -0
- llama_stack/apis/datasets/datasets.py +251 -0
- llama_stack/apis/datatypes.py +160 -0
- llama_stack/apis/eval/__init__.py +7 -0
- llama_stack/apis/eval/eval.py +169 -0
- llama_stack/apis/files/__init__.py +7 -0
- llama_stack/apis/files/files.py +199 -0
- llama_stack/apis/inference/__init__.py +1 -1
- llama_stack/apis/inference/inference.py +1169 -113
- llama_stack/apis/inspect/__init__.py +1 -1
- llama_stack/apis/inspect/inspect.py +69 -16
- llama_stack/apis/models/__init__.py +1 -1
- llama_stack/apis/models/models.py +148 -21
- llama_stack/apis/post_training/__init__.py +1 -1
- llama_stack/apis/post_training/post_training.py +265 -120
- llama_stack/{providers/adapters/agents/sample/config.py → apis/prompts/__init__.py} +2 -5
- llama_stack/apis/prompts/prompts.py +204 -0
- llama_stack/apis/providers/__init__.py +7 -0
- llama_stack/apis/providers/providers.py +69 -0
- llama_stack/apis/resource.py +37 -0
- llama_stack/apis/safety/__init__.py +1 -1
- llama_stack/apis/safety/safety.py +95 -12
- llama_stack/apis/scoring/__init__.py +7 -0
- llama_stack/apis/scoring/scoring.py +93 -0
- llama_stack/apis/scoring_functions/__init__.py +7 -0
- llama_stack/apis/scoring_functions/scoring_functions.py +208 -0
- llama_stack/apis/shields/__init__.py +1 -1
- llama_stack/apis/shields/shields.py +76 -33
- llama_stack/apis/synthetic_data_generation/__init__.py +1 -1
- llama_stack/apis/synthetic_data_generation/synthetic_data_generation.py +40 -17
- llama_stack/apis/telemetry/__init__.py +1 -1
- llama_stack/apis/telemetry/telemetry.py +322 -31
- llama_stack/apis/{dataset → tools}/__init__.py +2 -1
- llama_stack/apis/tools/rag_tool.py +218 -0
- llama_stack/apis/tools/tools.py +221 -0
- llama_stack/apis/vector_io/__init__.py +7 -0
- llama_stack/apis/vector_io/vector_io.py +960 -0
- llama_stack/apis/vector_stores/__init__.py +7 -0
- llama_stack/apis/vector_stores/vector_stores.py +51 -0
- llama_stack/apis/version.py +9 -0
- llama_stack/cli/llama.py +13 -5
- llama_stack/cli/stack/_list_deps.py +182 -0
- llama_stack/cli/stack/list_apis.py +1 -1
- llama_stack/cli/stack/list_deps.py +55 -0
- llama_stack/cli/stack/list_providers.py +24 -10
- llama_stack/cli/stack/list_stacks.py +56 -0
- llama_stack/cli/stack/remove.py +115 -0
- llama_stack/cli/stack/run.py +169 -56
- llama_stack/cli/stack/stack.py +18 -4
- llama_stack/cli/stack/utils.py +151 -0
- llama_stack/cli/table.py +23 -61
- llama_stack/cli/utils.py +29 -0
- llama_stack/core/access_control/access_control.py +131 -0
- llama_stack/core/access_control/conditions.py +129 -0
- llama_stack/core/access_control/datatypes.py +107 -0
- llama_stack/core/build.py +164 -0
- llama_stack/core/client.py +205 -0
- llama_stack/core/common.sh +37 -0
- llama_stack/{distribution → core}/configure.py +74 -55
- llama_stack/core/conversations/conversations.py +309 -0
- llama_stack/core/datatypes.py +625 -0
- llama_stack/core/distribution.py +276 -0
- llama_stack/core/external.py +54 -0
- llama_stack/core/id_generation.py +42 -0
- llama_stack/core/inspect.py +86 -0
- llama_stack/core/library_client.py +539 -0
- llama_stack/core/prompts/prompts.py +234 -0
- llama_stack/core/providers.py +137 -0
- llama_stack/core/request_headers.py +115 -0
- llama_stack/core/resolver.py +506 -0
- llama_stack/core/routers/__init__.py +101 -0
- llama_stack/core/routers/datasets.py +73 -0
- llama_stack/core/routers/eval_scoring.py +155 -0
- llama_stack/core/routers/inference.py +645 -0
- llama_stack/core/routers/safety.py +85 -0
- llama_stack/core/routers/tool_runtime.py +91 -0
- llama_stack/core/routers/vector_io.py +442 -0
- llama_stack/core/routing_tables/benchmarks.py +62 -0
- llama_stack/core/routing_tables/common.py +254 -0
- llama_stack/core/routing_tables/datasets.py +91 -0
- llama_stack/core/routing_tables/models.py +163 -0
- llama_stack/core/routing_tables/scoring_functions.py +66 -0
- llama_stack/core/routing_tables/shields.py +61 -0
- llama_stack/core/routing_tables/toolgroups.py +129 -0
- llama_stack/core/routing_tables/vector_stores.py +292 -0
- llama_stack/core/server/auth.py +187 -0
- llama_stack/core/server/auth_providers.py +494 -0
- llama_stack/core/server/quota.py +110 -0
- llama_stack/core/server/routes.py +141 -0
- llama_stack/core/server/server.py +542 -0
- llama_stack/core/server/tracing.py +80 -0
- llama_stack/core/stack.py +546 -0
- llama_stack/core/start_stack.sh +117 -0
- llama_stack/core/storage/datatypes.py +283 -0
- llama_stack/{cli/model → core/store}/__init__.py +1 -1
- llama_stack/core/store/registry.py +199 -0
- llama_stack/core/testing_context.py +49 -0
- llama_stack/core/ui/app.py +55 -0
- llama_stack/core/ui/modules/api.py +32 -0
- llama_stack/core/ui/modules/utils.py +42 -0
- llama_stack/core/ui/page/distribution/datasets.py +18 -0
- llama_stack/core/ui/page/distribution/eval_tasks.py +20 -0
- llama_stack/core/ui/page/distribution/models.py +18 -0
- llama_stack/core/ui/page/distribution/providers.py +27 -0
- llama_stack/core/ui/page/distribution/resources.py +48 -0
- llama_stack/core/ui/page/distribution/scoring_functions.py +18 -0
- llama_stack/core/ui/page/distribution/shields.py +19 -0
- llama_stack/core/ui/page/evaluations/app_eval.py +143 -0
- llama_stack/core/ui/page/evaluations/native_eval.py +253 -0
- llama_stack/core/ui/page/playground/chat.py +130 -0
- llama_stack/core/ui/page/playground/tools.py +352 -0
- llama_stack/core/utils/config.py +30 -0
- llama_stack/{distribution → core}/utils/config_dirs.py +3 -6
- llama_stack/core/utils/config_resolution.py +125 -0
- llama_stack/core/utils/context.py +84 -0
- llama_stack/core/utils/exec.py +96 -0
- llama_stack/{providers/impls/meta_reference/codeshield/config.py → core/utils/image_types.py} +4 -3
- llama_stack/{distribution → core}/utils/model_utils.py +2 -2
- llama_stack/{distribution → core}/utils/prompt_for_config.py +30 -63
- llama_stack/{apis/batch_inference → distributions/dell}/__init__.py +1 -1
- llama_stack/distributions/dell/build.yaml +33 -0
- llama_stack/distributions/dell/dell.py +158 -0
- llama_stack/distributions/dell/run-with-safety.yaml +141 -0
- llama_stack/distributions/dell/run.yaml +132 -0
- llama_stack/distributions/meta-reference-gpu/__init__.py +7 -0
- llama_stack/distributions/meta-reference-gpu/build.yaml +32 -0
- llama_stack/distributions/meta-reference-gpu/meta_reference.py +163 -0
- llama_stack/distributions/meta-reference-gpu/run-with-safety.yaml +154 -0
- llama_stack/distributions/meta-reference-gpu/run.yaml +139 -0
- llama_stack/{apis/evals → distributions/nvidia}/__init__.py +1 -1
- llama_stack/distributions/nvidia/build.yaml +29 -0
- llama_stack/distributions/nvidia/nvidia.py +154 -0
- llama_stack/distributions/nvidia/run-with-safety.yaml +137 -0
- llama_stack/distributions/nvidia/run.yaml +116 -0
- llama_stack/distributions/open-benchmark/__init__.py +7 -0
- llama_stack/distributions/open-benchmark/build.yaml +36 -0
- llama_stack/distributions/open-benchmark/open_benchmark.py +303 -0
- llama_stack/distributions/open-benchmark/run.yaml +252 -0
- llama_stack/distributions/postgres-demo/__init__.py +7 -0
- llama_stack/distributions/postgres-demo/build.yaml +23 -0
- llama_stack/distributions/postgres-demo/postgres_demo.py +125 -0
- llama_stack/distributions/postgres-demo/run.yaml +115 -0
- llama_stack/{apis/memory → distributions/starter}/__init__.py +1 -1
- llama_stack/distributions/starter/build.yaml +61 -0
- llama_stack/distributions/starter/run-with-postgres-store.yaml +285 -0
- llama_stack/distributions/starter/run.yaml +276 -0
- llama_stack/distributions/starter/starter.py +345 -0
- llama_stack/distributions/starter-gpu/__init__.py +7 -0
- llama_stack/distributions/starter-gpu/build.yaml +61 -0
- llama_stack/distributions/starter-gpu/run-with-postgres-store.yaml +288 -0
- llama_stack/distributions/starter-gpu/run.yaml +279 -0
- llama_stack/distributions/starter-gpu/starter_gpu.py +20 -0
- llama_stack/distributions/template.py +456 -0
- llama_stack/distributions/watsonx/__init__.py +7 -0
- llama_stack/distributions/watsonx/build.yaml +33 -0
- llama_stack/distributions/watsonx/run.yaml +133 -0
- llama_stack/distributions/watsonx/watsonx.py +95 -0
- llama_stack/env.py +24 -0
- llama_stack/log.py +314 -0
- llama_stack/models/llama/checkpoint.py +164 -0
- llama_stack/models/llama/datatypes.py +164 -0
- llama_stack/models/llama/hadamard_utils.py +86 -0
- llama_stack/models/llama/llama3/args.py +74 -0
- llama_stack/models/llama/llama3/chat_format.py +286 -0
- llama_stack/models/llama/llama3/generation.py +376 -0
- llama_stack/models/llama/llama3/interface.py +255 -0
- llama_stack/models/llama/llama3/model.py +304 -0
- llama_stack/models/llama/llama3/multimodal/__init__.py +12 -0
- llama_stack/models/llama/llama3/multimodal/encoder_utils.py +180 -0
- llama_stack/models/llama/llama3/multimodal/image_transform.py +409 -0
- llama_stack/models/llama/llama3/multimodal/model.py +1430 -0
- llama_stack/models/llama/llama3/multimodal/utils.py +26 -0
- llama_stack/models/llama/llama3/prompt_templates/__init__.py +22 -0
- llama_stack/models/llama/llama3/prompt_templates/base.py +39 -0
- llama_stack/models/llama/llama3/prompt_templates/system_prompts.py +319 -0
- llama_stack/models/llama/llama3/prompt_templates/tool_response.py +62 -0
- llama_stack/models/llama/llama3/quantization/loader.py +316 -0
- llama_stack/models/llama/llama3/template_data.py +116 -0
- llama_stack/models/llama/llama3/tokenizer.model +128000 -0
- llama_stack/models/llama/llama3/tokenizer.py +198 -0
- llama_stack/models/llama/llama3/tool_utils.py +266 -0
- llama_stack/models/llama/llama3_1/__init__.py +12 -0
- llama_stack/models/llama/llama3_1/prompt_format.md +358 -0
- llama_stack/models/llama/llama3_1/prompts.py +258 -0
- llama_stack/models/llama/llama3_2/prompts_text.py +229 -0
- llama_stack/models/llama/llama3_2/prompts_vision.py +126 -0
- llama_stack/models/llama/llama3_2/text_prompt_format.md +286 -0
- llama_stack/models/llama/llama3_2/vision_prompt_format.md +141 -0
- llama_stack/models/llama/llama3_3/prompts.py +259 -0
- llama_stack/models/llama/llama4/args.py +107 -0
- llama_stack/models/llama/llama4/chat_format.py +317 -0
- llama_stack/models/llama/llama4/datatypes.py +56 -0
- llama_stack/models/llama/llama4/ffn.py +58 -0
- llama_stack/models/llama/llama4/generation.py +313 -0
- llama_stack/models/llama/llama4/model.py +437 -0
- llama_stack/models/llama/llama4/moe.py +214 -0
- llama_stack/models/llama/llama4/preprocess.py +435 -0
- llama_stack/models/llama/llama4/prompt_format.md +304 -0
- llama_stack/models/llama/llama4/prompt_templates/system_prompts.py +136 -0
- llama_stack/models/llama/llama4/prompts.py +279 -0
- llama_stack/models/llama/llama4/quantization/__init__.py +5 -0
- llama_stack/models/llama/llama4/quantization/loader.py +226 -0
- llama_stack/models/llama/llama4/tokenizer.model +200000 -0
- llama_stack/models/llama/llama4/tokenizer.py +263 -0
- llama_stack/models/llama/llama4/vision/__init__.py +5 -0
- llama_stack/models/llama/llama4/vision/embedding.py +210 -0
- llama_stack/models/llama/llama4/vision/encoder.py +412 -0
- llama_stack/models/llama/prompt_format.py +191 -0
- llama_stack/models/llama/quantize_impls.py +316 -0
- llama_stack/models/llama/sku_list.py +1029 -0
- llama_stack/models/llama/sku_types.py +233 -0
- llama_stack/models/llama/tokenizer_utils.py +40 -0
- llama_stack/providers/datatypes.py +136 -107
- llama_stack/providers/inline/__init__.py +5 -0
- llama_stack/providers/inline/agents/__init__.py +5 -0
- llama_stack/providers/{impls/meta_reference/agents → inline/agents/meta_reference}/__init__.py +12 -5
- llama_stack/providers/inline/agents/meta_reference/agent_instance.py +1024 -0
- llama_stack/providers/inline/agents/meta_reference/agents.py +383 -0
- llama_stack/providers/inline/agents/meta_reference/config.py +37 -0
- llama_stack/providers/inline/agents/meta_reference/persistence.py +228 -0
- llama_stack/providers/inline/agents/meta_reference/responses/__init__.py +5 -0
- llama_stack/providers/inline/agents/meta_reference/responses/openai_responses.py +423 -0
- llama_stack/providers/inline/agents/meta_reference/responses/streaming.py +1226 -0
- llama_stack/providers/inline/agents/meta_reference/responses/tool_executor.py +449 -0
- llama_stack/providers/inline/agents/meta_reference/responses/types.py +194 -0
- llama_stack/providers/inline/agents/meta_reference/responses/utils.py +365 -0
- llama_stack/providers/inline/agents/meta_reference/safety.py +52 -0
- llama_stack/providers/inline/batches/__init__.py +5 -0
- llama_stack/providers/inline/batches/reference/__init__.py +36 -0
- llama_stack/providers/inline/batches/reference/batches.py +679 -0
- llama_stack/providers/inline/batches/reference/config.py +40 -0
- llama_stack/providers/inline/datasetio/__init__.py +5 -0
- llama_stack/providers/inline/datasetio/localfs/__init__.py +20 -0
- llama_stack/providers/inline/datasetio/localfs/config.py +23 -0
- llama_stack/providers/inline/datasetio/localfs/datasetio.py +113 -0
- llama_stack/providers/inline/eval/__init__.py +5 -0
- llama_stack/providers/inline/eval/meta_reference/__init__.py +28 -0
- llama_stack/providers/inline/eval/meta_reference/config.py +23 -0
- llama_stack/providers/inline/eval/meta_reference/eval.py +259 -0
- llama_stack/providers/inline/files/localfs/__init__.py +20 -0
- llama_stack/providers/inline/files/localfs/config.py +31 -0
- llama_stack/providers/inline/files/localfs/files.py +219 -0
- llama_stack/providers/inline/inference/__init__.py +5 -0
- llama_stack/providers/{impls/meta_reference/inference → inline/inference/meta_reference}/__init__.py +4 -4
- llama_stack/providers/inline/inference/meta_reference/common.py +24 -0
- llama_stack/providers/inline/inference/meta_reference/config.py +68 -0
- llama_stack/providers/inline/inference/meta_reference/generators.py +211 -0
- llama_stack/providers/inline/inference/meta_reference/inference.py +158 -0
- llama_stack/providers/inline/inference/meta_reference/model_parallel.py +96 -0
- llama_stack/providers/{impls/meta_reference/inference → inline/inference/meta_reference}/parallel_utils.py +56 -73
- llama_stack/providers/inline/inference/sentence_transformers/__init__.py +22 -0
- llama_stack/providers/{impls/meta_reference/agents → inline/inference/sentence_transformers}/config.py +6 -4
- llama_stack/providers/inline/inference/sentence_transformers/sentence_transformers.py +83 -0
- llama_stack/providers/inline/post_training/__init__.py +5 -0
- llama_stack/providers/inline/post_training/common/__init__.py +5 -0
- llama_stack/providers/inline/post_training/common/utils.py +35 -0
- llama_stack/providers/inline/post_training/common/validator.py +36 -0
- llama_stack/providers/inline/post_training/huggingface/__init__.py +27 -0
- llama_stack/providers/inline/post_training/huggingface/config.py +83 -0
- llama_stack/providers/inline/post_training/huggingface/post_training.py +208 -0
- llama_stack/providers/inline/post_training/huggingface/recipes/__init__.py +5 -0
- llama_stack/providers/inline/post_training/huggingface/recipes/finetune_single_device.py +519 -0
- llama_stack/providers/inline/post_training/huggingface/recipes/finetune_single_device_dpo.py +485 -0
- llama_stack/providers/inline/post_training/huggingface/utils.py +269 -0
- llama_stack/providers/inline/post_training/torchtune/__init__.py +27 -0
- llama_stack/providers/inline/post_training/torchtune/common/__init__.py +5 -0
- llama_stack/providers/inline/post_training/torchtune/common/checkpointer.py +240 -0
- llama_stack/providers/inline/post_training/torchtune/common/utils.py +99 -0
- llama_stack/providers/inline/post_training/torchtune/config.py +20 -0
- llama_stack/providers/inline/post_training/torchtune/datasets/__init__.py +5 -0
- llama_stack/providers/inline/post_training/torchtune/datasets/format_adapter.py +57 -0
- llama_stack/providers/inline/post_training/torchtune/datasets/sft.py +78 -0
- llama_stack/providers/inline/post_training/torchtune/post_training.py +178 -0
- llama_stack/providers/inline/post_training/torchtune/recipes/__init__.py +5 -0
- llama_stack/providers/inline/post_training/torchtune/recipes/lora_finetuning_single_device.py +588 -0
- llama_stack/providers/inline/safety/__init__.py +5 -0
- llama_stack/providers/{impls/meta_reference/codeshield → inline/safety/code_scanner}/__init__.py +4 -2
- llama_stack/providers/inline/safety/code_scanner/code_scanner.py +128 -0
- llama_stack/providers/{impls/meta_reference/memory → inline/safety/code_scanner}/config.py +5 -3
- llama_stack/providers/inline/safety/llama_guard/__init__.py +19 -0
- llama_stack/providers/inline/safety/llama_guard/config.py +19 -0
- llama_stack/providers/inline/safety/llama_guard/llama_guard.py +489 -0
- llama_stack/providers/{adapters/memory/sample → inline/safety/prompt_guard}/__init__.py +4 -4
- llama_stack/providers/inline/safety/prompt_guard/config.py +32 -0
- llama_stack/providers/inline/safety/prompt_guard/prompt_guard.py +131 -0
- llama_stack/providers/inline/scoring/__init__.py +5 -0
- llama_stack/providers/inline/scoring/basic/__init__.py +25 -0
- llama_stack/providers/{adapters/memory/weaviate → inline/scoring/basic}/config.py +5 -7
- llama_stack/providers/inline/scoring/basic/scoring.py +126 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/__init__.py +5 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/docvqa_scoring_fn.py +240 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/equality_scoring_fn.py +41 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/fn_defs/__init__.py +5 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/fn_defs/docvqa.py +21 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/fn_defs/equality.py +21 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/fn_defs/ifeval.py +23 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/fn_defs/regex_parser_math_response.py +27 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/fn_defs/regex_parser_multiple_choice_answer.py +71 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/fn_defs/subset_of.py +21 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/ifeval_scoring_fn.py +80 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/regex_parser_math_response_scoring_fn.py +66 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/regex_parser_scoring_fn.py +58 -0
- llama_stack/providers/inline/scoring/basic/scoring_fn/subset_of_scoring_fn.py +38 -0
- llama_stack/providers/inline/scoring/basic/utils/__init__.py +5 -0
- llama_stack/providers/inline/scoring/basic/utils/ifeval_utils.py +3319 -0
- llama_stack/providers/inline/scoring/basic/utils/math_utils.py +330 -0
- llama_stack/providers/inline/scoring/braintrust/__init__.py +27 -0
- llama_stack/providers/inline/scoring/braintrust/braintrust.py +230 -0
- llama_stack/providers/inline/scoring/braintrust/config.py +21 -0
- llama_stack/providers/inline/scoring/braintrust/scoring_fn/__init__.py +5 -0
- llama_stack/providers/inline/scoring/braintrust/scoring_fn/fn_defs/__init__.py +5 -0
- llama_stack/providers/inline/scoring/braintrust/scoring_fn/fn_defs/answer_correctness.py +24 -0
- llama_stack/providers/inline/scoring/braintrust/scoring_fn/fn_defs/answer_relevancy.py +24 -0
- llama_stack/providers/inline/scoring/braintrust/scoring_fn/fn_defs/answer_similarity.py +24 -0
- llama_stack/providers/inline/scoring/braintrust/scoring_fn/fn_defs/context_entity_recall.py +24 -0
- llama_stack/providers/inline/scoring/braintrust/scoring_fn/fn_defs/context_precision.py +24 -0
- llama_stack/providers/inline/scoring/braintrust/scoring_fn/fn_defs/context_recall.py +24 -0
- llama_stack/providers/inline/scoring/braintrust/scoring_fn/fn_defs/context_relevancy.py +23 -0
- llama_stack/providers/inline/scoring/braintrust/scoring_fn/fn_defs/factuality.py +24 -0
- llama_stack/providers/inline/scoring/braintrust/scoring_fn/fn_defs/faithfulness.py +24 -0
- llama_stack/providers/inline/scoring/llm_as_judge/__init__.py +21 -0
- llama_stack/providers/inline/scoring/llm_as_judge/config.py +14 -0
- llama_stack/providers/inline/scoring/llm_as_judge/scoring.py +113 -0
- llama_stack/providers/inline/scoring/llm_as_judge/scoring_fn/__init__.py +5 -0
- llama_stack/providers/inline/scoring/llm_as_judge/scoring_fn/fn_defs/__init__.py +5 -0
- llama_stack/providers/inline/scoring/llm_as_judge/scoring_fn/fn_defs/llm_as_judge_405b_simpleqa.py +96 -0
- llama_stack/providers/inline/scoring/llm_as_judge/scoring_fn/fn_defs/llm_as_judge_base.py +20 -0
- llama_stack/providers/inline/scoring/llm_as_judge/scoring_fn/llm_as_judge_scoring_fn.py +81 -0
- llama_stack/providers/inline/telemetry/__init__.py +5 -0
- llama_stack/providers/inline/telemetry/meta_reference/__init__.py +21 -0
- llama_stack/providers/inline/telemetry/meta_reference/config.py +47 -0
- llama_stack/providers/inline/telemetry/meta_reference/telemetry.py +252 -0
- llama_stack/providers/inline/tool_runtime/__init__.py +5 -0
- llama_stack/providers/inline/tool_runtime/rag/__init__.py +19 -0
- llama_stack/providers/{impls/meta_reference/telemetry → inline/tool_runtime/rag}/config.py +5 -3
- llama_stack/providers/inline/tool_runtime/rag/context_retriever.py +77 -0
- llama_stack/providers/inline/tool_runtime/rag/memory.py +332 -0
- llama_stack/providers/inline/vector_io/__init__.py +5 -0
- llama_stack/providers/inline/vector_io/chroma/__init__.py +19 -0
- llama_stack/providers/inline/vector_io/chroma/config.py +30 -0
- llama_stack/providers/inline/vector_io/faiss/__init__.py +21 -0
- llama_stack/providers/inline/vector_io/faiss/config.py +26 -0
- llama_stack/providers/inline/vector_io/faiss/faiss.py +293 -0
- llama_stack/providers/inline/vector_io/milvus/__init__.py +19 -0
- llama_stack/providers/inline/vector_io/milvus/config.py +29 -0
- llama_stack/providers/inline/vector_io/qdrant/__init__.py +20 -0
- llama_stack/providers/inline/vector_io/qdrant/config.py +29 -0
- llama_stack/providers/inline/vector_io/sqlite_vec/__init__.py +20 -0
- llama_stack/providers/inline/vector_io/sqlite_vec/config.py +26 -0
- llama_stack/providers/inline/vector_io/sqlite_vec/sqlite_vec.py +483 -0
- llama_stack/providers/registry/agents.py +16 -18
- llama_stack/providers/registry/batches.py +26 -0
- llama_stack/providers/registry/datasetio.py +49 -0
- llama_stack/providers/registry/eval.py +46 -0
- llama_stack/providers/registry/files.py +31 -0
- llama_stack/providers/registry/inference.py +273 -118
- llama_stack/providers/registry/post_training.py +69 -0
- llama_stack/providers/registry/safety.py +46 -41
- llama_stack/providers/registry/scoring.py +51 -0
- llama_stack/providers/registry/tool_runtime.py +87 -0
- llama_stack/providers/registry/vector_io.py +828 -0
- llama_stack/providers/remote/__init__.py +5 -0
- llama_stack/providers/remote/agents/__init__.py +5 -0
- llama_stack/providers/remote/datasetio/__init__.py +5 -0
- llama_stack/providers/{adapters/memory/chroma → remote/datasetio/huggingface}/__init__.py +7 -4
- llama_stack/providers/remote/datasetio/huggingface/config.py +23 -0
- llama_stack/providers/remote/datasetio/huggingface/huggingface.py +99 -0
- llama_stack/providers/remote/datasetio/nvidia/__init__.py +23 -0
- llama_stack/providers/remote/datasetio/nvidia/config.py +61 -0
- llama_stack/providers/remote/datasetio/nvidia/datasetio.py +116 -0
- llama_stack/providers/remote/eval/__init__.py +5 -0
- llama_stack/providers/remote/eval/nvidia/__init__.py +31 -0
- llama_stack/providers/remote/eval/nvidia/config.py +29 -0
- llama_stack/providers/remote/eval/nvidia/eval.py +162 -0
- llama_stack/providers/remote/files/s3/__init__.py +19 -0
- llama_stack/providers/remote/files/s3/config.py +42 -0
- llama_stack/providers/remote/files/s3/files.py +313 -0
- llama_stack/providers/remote/inference/__init__.py +5 -0
- llama_stack/providers/{adapters/safety/sample → remote/inference/anthropic}/__init__.py +4 -6
- llama_stack/providers/remote/inference/anthropic/anthropic.py +36 -0
- llama_stack/providers/remote/inference/anthropic/config.py +28 -0
- llama_stack/providers/{impls/meta_reference/telemetry → remote/inference/azure}/__init__.py +4 -4
- llama_stack/providers/remote/inference/azure/azure.py +25 -0
- llama_stack/providers/remote/inference/azure/config.py +61 -0
- llama_stack/providers/{adapters → remote}/inference/bedrock/__init__.py +18 -17
- llama_stack/providers/remote/inference/bedrock/bedrock.py +142 -0
- llama_stack/providers/{adapters/inference/sample → remote/inference/bedrock}/config.py +3 -4
- llama_stack/providers/remote/inference/bedrock/models.py +29 -0
- llama_stack/providers/remote/inference/cerebras/__init__.py +19 -0
- llama_stack/providers/remote/inference/cerebras/cerebras.py +28 -0
- llama_stack/providers/remote/inference/cerebras/config.py +30 -0
- llama_stack/providers/{adapters → remote}/inference/databricks/__init__.py +4 -5
- llama_stack/providers/remote/inference/databricks/config.py +37 -0
- llama_stack/providers/remote/inference/databricks/databricks.py +44 -0
- llama_stack/providers/{adapters → remote}/inference/fireworks/__init__.py +8 -4
- llama_stack/providers/remote/inference/fireworks/config.py +27 -0
- llama_stack/providers/remote/inference/fireworks/fireworks.py +27 -0
- llama_stack/providers/{adapters/memory/pgvector → remote/inference/gemini}/__init__.py +4 -4
- llama_stack/providers/remote/inference/gemini/config.py +28 -0
- llama_stack/providers/remote/inference/gemini/gemini.py +82 -0
- llama_stack/providers/remote/inference/groq/__init__.py +15 -0
- llama_stack/providers/remote/inference/groq/config.py +34 -0
- llama_stack/providers/remote/inference/groq/groq.py +18 -0
- llama_stack/providers/remote/inference/llama_openai_compat/__init__.py +15 -0
- llama_stack/providers/remote/inference/llama_openai_compat/config.py +34 -0
- llama_stack/providers/remote/inference/llama_openai_compat/llama.py +46 -0
- llama_stack/providers/remote/inference/nvidia/__init__.py +23 -0
- llama_stack/providers/remote/inference/nvidia/config.py +64 -0
- llama_stack/providers/remote/inference/nvidia/nvidia.py +61 -0
- llama_stack/providers/{adapters/safety/sample/config.py → remote/inference/nvidia/utils.py} +3 -4
- llama_stack/providers/{impls/vllm → remote/inference/ollama}/__init__.py +4 -6
- llama_stack/providers/remote/inference/ollama/config.py +25 -0
- llama_stack/providers/remote/inference/ollama/ollama.py +102 -0
- llama_stack/providers/{adapters/telemetry/opentelemetry → remote/inference/openai}/__init__.py +4 -4
- llama_stack/providers/remote/inference/openai/config.py +39 -0
- llama_stack/providers/remote/inference/openai/openai.py +38 -0
- llama_stack/providers/remote/inference/passthrough/__init__.py +23 -0
- llama_stack/providers/remote/inference/passthrough/config.py +34 -0
- llama_stack/providers/remote/inference/passthrough/passthrough.py +122 -0
- llama_stack/providers/remote/inference/runpod/__init__.py +16 -0
- llama_stack/providers/remote/inference/runpod/config.py +32 -0
- llama_stack/providers/remote/inference/runpod/runpod.py +42 -0
- llama_stack/providers/remote/inference/sambanova/__init__.py +16 -0
- llama_stack/providers/remote/inference/sambanova/config.py +34 -0
- llama_stack/providers/remote/inference/sambanova/sambanova.py +28 -0
- llama_stack/providers/{adapters → remote}/inference/tgi/__init__.py +3 -4
- llama_stack/providers/remote/inference/tgi/config.py +76 -0
- llama_stack/providers/remote/inference/tgi/tgi.py +85 -0
- llama_stack/providers/{adapters → remote}/inference/together/__init__.py +8 -4
- llama_stack/providers/remote/inference/together/config.py +27 -0
- llama_stack/providers/remote/inference/together/together.py +102 -0
- llama_stack/providers/remote/inference/vertexai/__init__.py +15 -0
- llama_stack/providers/remote/inference/vertexai/config.py +48 -0
- llama_stack/providers/remote/inference/vertexai/vertexai.py +54 -0
- llama_stack/providers/remote/inference/vllm/__init__.py +22 -0
- llama_stack/providers/remote/inference/vllm/config.py +59 -0
- llama_stack/providers/remote/inference/vllm/vllm.py +111 -0
- llama_stack/providers/remote/inference/watsonx/__init__.py +15 -0
- llama_stack/providers/remote/inference/watsonx/config.py +45 -0
- llama_stack/providers/remote/inference/watsonx/watsonx.py +336 -0
- llama_stack/providers/remote/post_training/__init__.py +5 -0
- llama_stack/providers/remote/post_training/nvidia/__init__.py +23 -0
- llama_stack/providers/remote/post_training/nvidia/config.py +113 -0
- llama_stack/providers/remote/post_training/nvidia/models.py +27 -0
- llama_stack/providers/remote/post_training/nvidia/post_training.py +430 -0
- llama_stack/providers/remote/post_training/nvidia/utils.py +63 -0
- llama_stack/providers/remote/safety/__init__.py +5 -0
- llama_stack/providers/remote/safety/bedrock/bedrock.py +111 -0
- llama_stack/providers/remote/safety/bedrock/config.py +14 -0
- llama_stack/providers/{adapters/inference/sample → remote/safety/nvidia}/__init__.py +5 -4
- llama_stack/providers/remote/safety/nvidia/config.py +40 -0
- llama_stack/providers/remote/safety/nvidia/nvidia.py +161 -0
- llama_stack/providers/{adapters/agents/sample → remote/safety/sambanova}/__init__.py +5 -4
- llama_stack/providers/remote/safety/sambanova/config.py +37 -0
- llama_stack/providers/remote/safety/sambanova/sambanova.py +98 -0
- llama_stack/providers/remote/tool_runtime/__init__.py +5 -0
- llama_stack/providers/remote/tool_runtime/bing_search/__init__.py +21 -0
- llama_stack/providers/remote/tool_runtime/bing_search/bing_search.py +112 -0
- llama_stack/providers/remote/tool_runtime/bing_search/config.py +22 -0
- llama_stack/providers/remote/tool_runtime/brave_search/__init__.py +20 -0
- llama_stack/providers/remote/tool_runtime/brave_search/brave_search.py +148 -0
- llama_stack/providers/remote/tool_runtime/brave_search/config.py +27 -0
- llama_stack/providers/remote/tool_runtime/model_context_protocol/__init__.py +15 -0
- llama_stack/providers/remote/tool_runtime/model_context_protocol/config.py +20 -0
- llama_stack/providers/remote/tool_runtime/model_context_protocol/model_context_protocol.py +73 -0
- llama_stack/providers/remote/tool_runtime/tavily_search/__init__.py +20 -0
- llama_stack/providers/remote/tool_runtime/tavily_search/config.py +27 -0
- llama_stack/providers/remote/tool_runtime/tavily_search/tavily_search.py +84 -0
- llama_stack/providers/remote/tool_runtime/wolfram_alpha/__init__.py +22 -0
- llama_stack/providers/remote/tool_runtime/wolfram_alpha/config.py +21 -0
- llama_stack/providers/remote/tool_runtime/wolfram_alpha/wolfram_alpha.py +140 -0
- llama_stack/providers/remote/vector_io/__init__.py +5 -0
- llama_stack/providers/remote/vector_io/chroma/__init__.py +17 -0
- llama_stack/providers/remote/vector_io/chroma/chroma.py +215 -0
- llama_stack/providers/remote/vector_io/chroma/config.py +28 -0
- llama_stack/providers/remote/vector_io/milvus/__init__.py +18 -0
- llama_stack/providers/remote/vector_io/milvus/config.py +35 -0
- llama_stack/providers/remote/vector_io/milvus/milvus.py +375 -0
- llama_stack/providers/remote/vector_io/pgvector/__init__.py +17 -0
- llama_stack/providers/remote/vector_io/pgvector/config.py +47 -0
- llama_stack/providers/remote/vector_io/pgvector/pgvector.py +460 -0
- llama_stack/providers/remote/vector_io/qdrant/__init__.py +17 -0
- llama_stack/providers/remote/vector_io/qdrant/config.py +37 -0
- llama_stack/providers/remote/vector_io/qdrant/qdrant.py +265 -0
- llama_stack/providers/remote/vector_io/weaviate/__init__.py +17 -0
- llama_stack/providers/remote/vector_io/weaviate/config.py +32 -0
- llama_stack/providers/remote/vector_io/weaviate/weaviate.py +393 -0
- llama_stack/providers/utils/bedrock/__init__.py +5 -0
- llama_stack/providers/utils/bedrock/client.py +74 -0
- llama_stack/providers/utils/bedrock/config.py +64 -0
- llama_stack/providers/utils/bedrock/refreshable_boto_session.py +112 -0
- llama_stack/providers/utils/common/__init__.py +5 -0
- llama_stack/providers/utils/common/data_schema_validator.py +103 -0
- llama_stack/providers/utils/datasetio/__init__.py +5 -0
- llama_stack/providers/utils/datasetio/url_utils.py +47 -0
- llama_stack/providers/utils/files/__init__.py +5 -0
- llama_stack/providers/utils/files/form_data.py +69 -0
- llama_stack/providers/utils/inference/__init__.py +8 -7
- llama_stack/providers/utils/inference/embedding_mixin.py +101 -0
- llama_stack/providers/utils/inference/inference_store.py +264 -0
- llama_stack/providers/utils/inference/litellm_openai_mixin.py +336 -0
- llama_stack/providers/utils/inference/model_registry.py +173 -23
- llama_stack/providers/utils/inference/openai_compat.py +1261 -49
- llama_stack/providers/utils/inference/openai_mixin.py +506 -0
- llama_stack/providers/utils/inference/prompt_adapter.py +365 -67
- llama_stack/providers/utils/kvstore/api.py +6 -6
- llama_stack/providers/utils/kvstore/config.py +28 -48
- llama_stack/providers/utils/kvstore/kvstore.py +61 -15
- llama_stack/providers/utils/kvstore/mongodb/__init__.py +9 -0
- llama_stack/providers/utils/kvstore/mongodb/mongodb.py +82 -0
- llama_stack/providers/utils/kvstore/postgres/__init__.py +7 -0
- llama_stack/providers/utils/kvstore/postgres/postgres.py +114 -0
- llama_stack/providers/utils/kvstore/redis/redis.py +33 -9
- llama_stack/providers/utils/kvstore/sqlite/config.py +2 -1
- llama_stack/providers/utils/kvstore/sqlite/sqlite.py +123 -22
- llama_stack/providers/utils/memory/file_utils.py +1 -1
- llama_stack/providers/utils/memory/openai_vector_store_mixin.py +1304 -0
- llama_stack/providers/utils/memory/vector_store.py +220 -82
- llama_stack/providers/utils/pagination.py +43 -0
- llama_stack/providers/utils/responses/__init__.py +5 -0
- llama_stack/providers/utils/responses/responses_store.py +292 -0
- llama_stack/providers/utils/scheduler.py +270 -0
- llama_stack/providers/utils/scoring/__init__.py +5 -0
- llama_stack/providers/utils/scoring/aggregation_utils.py +75 -0
- llama_stack/providers/utils/scoring/base_scoring_fn.py +114 -0
- llama_stack/providers/utils/scoring/basic_scoring_utils.py +26 -0
- llama_stack/providers/utils/sqlstore/__init__.py +5 -0
- llama_stack/providers/utils/sqlstore/api.py +128 -0
- llama_stack/providers/utils/sqlstore/authorized_sqlstore.py +319 -0
- llama_stack/providers/utils/sqlstore/sqlalchemy_sqlstore.py +343 -0
- llama_stack/providers/utils/sqlstore/sqlstore.py +70 -0
- llama_stack/providers/utils/telemetry/trace_protocol.py +142 -0
- llama_stack/providers/utils/telemetry/tracing.py +192 -53
- llama_stack/providers/utils/tools/__init__.py +5 -0
- llama_stack/providers/utils/tools/mcp.py +148 -0
- llama_stack/providers/utils/tools/ttl_dict.py +70 -0
- llama_stack/providers/utils/vector_io/__init__.py +5 -0
- llama_stack/providers/utils/vector_io/vector_utils.py +156 -0
- llama_stack/schema_utils.py +118 -0
- llama_stack/strong_typing/__init__.py +19 -0
- llama_stack/strong_typing/auxiliary.py +228 -0
- llama_stack/strong_typing/classdef.py +440 -0
- llama_stack/strong_typing/core.py +46 -0
- llama_stack/strong_typing/deserializer.py +877 -0
- llama_stack/strong_typing/docstring.py +409 -0
- llama_stack/strong_typing/exception.py +23 -0
- llama_stack/strong_typing/inspection.py +1085 -0
- llama_stack/strong_typing/mapping.py +40 -0
- llama_stack/strong_typing/name.py +182 -0
- llama_stack/strong_typing/py.typed +0 -0
- llama_stack/strong_typing/schema.py +792 -0
- llama_stack/strong_typing/serialization.py +97 -0
- llama_stack/strong_typing/serializer.py +500 -0
- llama_stack/strong_typing/slots.py +27 -0
- llama_stack/strong_typing/topological.py +89 -0
- llama_stack/testing/__init__.py +5 -0
- llama_stack/testing/api_recorder.py +956 -0
- llama_stack/ui/node_modules/flatted/python/flatted.py +149 -0
- llama_stack-0.3.4.dist-info/METADATA +261 -0
- llama_stack-0.3.4.dist-info/RECORD +625 -0
- {llama_stack-0.0.42.dist-info → llama_stack-0.3.4.dist-info}/WHEEL +1 -1
- llama_stack/apis/agents/client.py +0 -292
- llama_stack/apis/agents/event_logger.py +0 -184
- llama_stack/apis/batch_inference/batch_inference.py +0 -72
- llama_stack/apis/common/deployment_types.py +0 -31
- llama_stack/apis/dataset/dataset.py +0 -63
- llama_stack/apis/evals/evals.py +0 -122
- llama_stack/apis/inference/client.py +0 -197
- llama_stack/apis/inspect/client.py +0 -82
- llama_stack/apis/memory/client.py +0 -155
- llama_stack/apis/memory/memory.py +0 -65
- llama_stack/apis/memory_banks/__init__.py +0 -7
- llama_stack/apis/memory_banks/client.py +0 -101
- llama_stack/apis/memory_banks/memory_banks.py +0 -78
- llama_stack/apis/models/client.py +0 -83
- llama_stack/apis/reward_scoring/__init__.py +0 -7
- llama_stack/apis/reward_scoring/reward_scoring.py +0 -55
- llama_stack/apis/safety/client.py +0 -105
- llama_stack/apis/shields/client.py +0 -79
- llama_stack/cli/download.py +0 -340
- llama_stack/cli/model/describe.py +0 -82
- llama_stack/cli/model/download.py +0 -24
- llama_stack/cli/model/list.py +0 -62
- llama_stack/cli/model/model.py +0 -34
- llama_stack/cli/model/prompt_format.py +0 -112
- llama_stack/cli/model/safety_models.py +0 -52
- llama_stack/cli/stack/build.py +0 -299
- llama_stack/cli/stack/configure.py +0 -178
- llama_stack/distribution/build.py +0 -123
- llama_stack/distribution/build_conda_env.sh +0 -136
- llama_stack/distribution/build_container.sh +0 -142
- llama_stack/distribution/common.sh +0 -40
- llama_stack/distribution/configure_container.sh +0 -47
- llama_stack/distribution/datatypes.py +0 -139
- llama_stack/distribution/distribution.py +0 -58
- llama_stack/distribution/inspect.py +0 -67
- llama_stack/distribution/request_headers.py +0 -57
- llama_stack/distribution/resolver.py +0 -323
- llama_stack/distribution/routers/__init__.py +0 -48
- llama_stack/distribution/routers/routers.py +0 -158
- llama_stack/distribution/routers/routing_tables.py +0 -173
- llama_stack/distribution/server/endpoints.py +0 -48
- llama_stack/distribution/server/server.py +0 -343
- llama_stack/distribution/start_conda_env.sh +0 -42
- llama_stack/distribution/start_container.sh +0 -64
- llama_stack/distribution/templates/local-bedrock-conda-example-build.yaml +0 -10
- llama_stack/distribution/templates/local-build.yaml +0 -10
- llama_stack/distribution/templates/local-databricks-build.yaml +0 -10
- llama_stack/distribution/templates/local-fireworks-build.yaml +0 -10
- llama_stack/distribution/templates/local-hf-endpoint-build.yaml +0 -10
- llama_stack/distribution/templates/local-hf-serverless-build.yaml +0 -10
- llama_stack/distribution/templates/local-ollama-build.yaml +0 -10
- llama_stack/distribution/templates/local-tgi-build.yaml +0 -10
- llama_stack/distribution/templates/local-together-build.yaml +0 -10
- llama_stack/distribution/templates/local-vllm-build.yaml +0 -10
- llama_stack/distribution/utils/exec.py +0 -105
- llama_stack/providers/adapters/agents/sample/sample.py +0 -18
- llama_stack/providers/adapters/inference/bedrock/bedrock.py +0 -451
- llama_stack/providers/adapters/inference/bedrock/config.py +0 -55
- llama_stack/providers/adapters/inference/databricks/config.py +0 -21
- llama_stack/providers/adapters/inference/databricks/databricks.py +0 -125
- llama_stack/providers/adapters/inference/fireworks/config.py +0 -20
- llama_stack/providers/adapters/inference/fireworks/fireworks.py +0 -130
- llama_stack/providers/adapters/inference/ollama/__init__.py +0 -19
- llama_stack/providers/adapters/inference/ollama/ollama.py +0 -175
- llama_stack/providers/adapters/inference/sample/sample.py +0 -23
- llama_stack/providers/adapters/inference/tgi/config.py +0 -43
- llama_stack/providers/adapters/inference/tgi/tgi.py +0 -200
- llama_stack/providers/adapters/inference/together/config.py +0 -22
- llama_stack/providers/adapters/inference/together/together.py +0 -143
- llama_stack/providers/adapters/memory/chroma/chroma.py +0 -157
- llama_stack/providers/adapters/memory/pgvector/config.py +0 -17
- llama_stack/providers/adapters/memory/pgvector/pgvector.py +0 -211
- llama_stack/providers/adapters/memory/sample/sample.py +0 -23
- llama_stack/providers/adapters/memory/weaviate/__init__.py +0 -15
- llama_stack/providers/adapters/memory/weaviate/weaviate.py +0 -190
- llama_stack/providers/adapters/safety/bedrock/bedrock.py +0 -113
- llama_stack/providers/adapters/safety/bedrock/config.py +0 -16
- llama_stack/providers/adapters/safety/sample/sample.py +0 -23
- llama_stack/providers/adapters/safety/together/__init__.py +0 -18
- llama_stack/providers/adapters/safety/together/config.py +0 -26
- llama_stack/providers/adapters/safety/together/together.py +0 -101
- llama_stack/providers/adapters/telemetry/opentelemetry/config.py +0 -12
- llama_stack/providers/adapters/telemetry/opentelemetry/opentelemetry.py +0 -201
- llama_stack/providers/adapters/telemetry/sample/__init__.py +0 -17
- llama_stack/providers/adapters/telemetry/sample/config.py +0 -12
- llama_stack/providers/adapters/telemetry/sample/sample.py +0 -18
- llama_stack/providers/impls/meta_reference/agents/agent_instance.py +0 -844
- llama_stack/providers/impls/meta_reference/agents/agents.py +0 -161
- llama_stack/providers/impls/meta_reference/agents/persistence.py +0 -84
- llama_stack/providers/impls/meta_reference/agents/rag/context_retriever.py +0 -74
- llama_stack/providers/impls/meta_reference/agents/safety.py +0 -57
- llama_stack/providers/impls/meta_reference/agents/tests/code_execution.py +0 -93
- llama_stack/providers/impls/meta_reference/agents/tests/test_chat_agent.py +0 -305
- llama_stack/providers/impls/meta_reference/agents/tools/base.py +0 -20
- llama_stack/providers/impls/meta_reference/agents/tools/builtin.py +0 -375
- llama_stack/providers/impls/meta_reference/agents/tools/ipython_tool/code_env_prefix.py +0 -133
- llama_stack/providers/impls/meta_reference/agents/tools/ipython_tool/code_execution.py +0 -256
- llama_stack/providers/impls/meta_reference/agents/tools/ipython_tool/matplotlib_custom_backend.py +0 -87
- llama_stack/providers/impls/meta_reference/agents/tools/ipython_tool/utils.py +0 -21
- llama_stack/providers/impls/meta_reference/agents/tools/safety.py +0 -43
- llama_stack/providers/impls/meta_reference/codeshield/code_scanner.py +0 -58
- llama_stack/providers/impls/meta_reference/inference/config.py +0 -45
- llama_stack/providers/impls/meta_reference/inference/generation.py +0 -376
- llama_stack/providers/impls/meta_reference/inference/inference.py +0 -280
- llama_stack/providers/impls/meta_reference/inference/model_parallel.py +0 -99
- llama_stack/providers/impls/meta_reference/inference/quantization/fp8_impls.py +0 -184
- llama_stack/providers/impls/meta_reference/inference/quantization/fp8_txest_disabled.py +0 -76
- llama_stack/providers/impls/meta_reference/inference/quantization/loader.py +0 -97
- llama_stack/providers/impls/meta_reference/inference/quantization/scripts/quantize_checkpoint.py +0 -161
- llama_stack/providers/impls/meta_reference/memory/__init__.py +0 -19
- llama_stack/providers/impls/meta_reference/memory/faiss.py +0 -113
- llama_stack/providers/impls/meta_reference/safety/__init__.py +0 -17
- llama_stack/providers/impls/meta_reference/safety/base.py +0 -57
- llama_stack/providers/impls/meta_reference/safety/config.py +0 -48
- llama_stack/providers/impls/meta_reference/safety/llama_guard.py +0 -268
- llama_stack/providers/impls/meta_reference/safety/prompt_guard.py +0 -145
- llama_stack/providers/impls/meta_reference/safety/safety.py +0 -112
- llama_stack/providers/impls/meta_reference/telemetry/console.py +0 -89
- llama_stack/providers/impls/vllm/config.py +0 -35
- llama_stack/providers/impls/vllm/vllm.py +0 -241
- llama_stack/providers/registry/memory.py +0 -78
- llama_stack/providers/registry/telemetry.py +0 -44
- llama_stack/providers/tests/agents/test_agents.py +0 -210
- llama_stack/providers/tests/inference/test_inference.py +0 -257
- llama_stack/providers/tests/inference/test_prompt_adapter.py +0 -126
- llama_stack/providers/tests/memory/test_memory.py +0 -136
- llama_stack/providers/tests/resolver.py +0 -100
- llama_stack/providers/tests/safety/test_safety.py +0 -77
- llama_stack-0.0.42.dist-info/METADATA +0 -137
- llama_stack-0.0.42.dist-info/RECORD +0 -256
- /llama_stack/{distribution → core}/__init__.py +0 -0
- /llama_stack/{distribution/server → core/access_control}/__init__.py +0 -0
- /llama_stack/{distribution/utils → core/conversations}/__init__.py +0 -0
- /llama_stack/{providers/adapters → core/prompts}/__init__.py +0 -0
- /llama_stack/{providers/adapters/agents → core/routing_tables}/__init__.py +0 -0
- /llama_stack/{providers/adapters/inference → core/server}/__init__.py +0 -0
- /llama_stack/{providers/adapters/memory → core/storage}/__init__.py +0 -0
- /llama_stack/{providers/adapters/safety → core/ui}/__init__.py +0 -0
- /llama_stack/{providers/adapters/telemetry → core/ui/modules}/__init__.py +0 -0
- /llama_stack/{providers/impls → core/ui/page}/__init__.py +0 -0
- /llama_stack/{providers/impls/meta_reference → core/ui/page/distribution}/__init__.py +0 -0
- /llama_stack/{providers/impls/meta_reference/agents/rag → core/ui/page/evaluations}/__init__.py +0 -0
- /llama_stack/{providers/impls/meta_reference/agents/tests → core/ui/page/playground}/__init__.py +0 -0
- /llama_stack/{providers/impls/meta_reference/agents/tools → core/utils}/__init__.py +0 -0
- /llama_stack/{distribution → core}/utils/dynamic.py +0 -0
- /llama_stack/{distribution → core}/utils/serialize.py +0 -0
- /llama_stack/{providers/impls/meta_reference/agents/tools/ipython_tool → distributions}/__init__.py +0 -0
- /llama_stack/{providers/impls/meta_reference/inference/quantization → models}/__init__.py +0 -0
- /llama_stack/{providers/impls/meta_reference/inference/quantization/scripts → models/llama}/__init__.py +0 -0
- /llama_stack/{providers/tests → models/llama/llama3}/__init__.py +0 -0
- /llama_stack/{providers/tests/agents → models/llama/llama3/quantization}/__init__.py +0 -0
- /llama_stack/{providers/tests/inference → models/llama/llama3_2}/__init__.py +0 -0
- /llama_stack/{providers/tests/memory → models/llama/llama3_3}/__init__.py +0 -0
- /llama_stack/{providers/tests/safety → models/llama/llama4}/__init__.py +0 -0
- /llama_stack/{scripts → models/llama/llama4/prompt_templates}/__init__.py +0 -0
- /llama_stack/providers/{adapters → remote}/safety/bedrock/__init__.py +0 -0
- {llama_stack-0.0.42.dist-info → llama_stack-0.3.4.dist-info}/entry_points.txt +0 -0
- {llama_stack-0.0.42.dist-info → llama_stack-0.3.4.dist-info/licenses}/LICENSE +0 -0
- {llama_stack-0.0.42.dist-info → llama_stack-0.3.4.dist-info}/top_level.txt +0 -0
|
@@ -0,0 +1,117 @@
|
|
|
1
|
+
#!/usr/bin/env bash
|
|
2
|
+
|
|
3
|
+
# Copyright (c) Meta Platforms, Inc. and affiliates.
|
|
4
|
+
# All rights reserved.
|
|
5
|
+
#
|
|
6
|
+
# This source code is licensed under the terms described in the LICENSE file in
|
|
7
|
+
# the root directory of this source tree.
|
|
8
|
+
|
|
9
|
+
|
|
10
|
+
TEST_PYPI_VERSION=${TEST_PYPI_VERSION:-}
|
|
11
|
+
PYPI_VERSION=${PYPI_VERSION:-}
|
|
12
|
+
VIRTUAL_ENV=${VIRTUAL_ENV:-}
|
|
13
|
+
|
|
14
|
+
set -euo pipefail
|
|
15
|
+
|
|
16
|
+
RED='\033[0;31m'
|
|
17
|
+
GREEN='\033[0;32m'
|
|
18
|
+
NC='\033[0m' # No Color
|
|
19
|
+
|
|
20
|
+
error_handler() {
|
|
21
|
+
echo "Error occurred in script at line: ${1}" >&2
|
|
22
|
+
exit 1
|
|
23
|
+
}
|
|
24
|
+
|
|
25
|
+
trap 'error_handler ${LINENO}' ERR
|
|
26
|
+
|
|
27
|
+
if [ $# -lt 3 ]; then
|
|
28
|
+
echo "Usage: $0 <env_type> <env_path_or_name> <port> [--config <yaml_config>]"
|
|
29
|
+
exit 1
|
|
30
|
+
fi
|
|
31
|
+
|
|
32
|
+
env_type="$1"
|
|
33
|
+
shift
|
|
34
|
+
|
|
35
|
+
env_path_or_name="$1"
|
|
36
|
+
container_image="localhost/$env_path_or_name"
|
|
37
|
+
shift
|
|
38
|
+
|
|
39
|
+
port="$1"
|
|
40
|
+
shift
|
|
41
|
+
|
|
42
|
+
SCRIPT_DIR=$(dirname "$(readlink -f "$0")")
|
|
43
|
+
|
|
44
|
+
# Initialize variables
|
|
45
|
+
yaml_config=""
|
|
46
|
+
other_args=""
|
|
47
|
+
|
|
48
|
+
# Process remaining arguments
|
|
49
|
+
while [[ $# -gt 0 ]]; do
|
|
50
|
+
case "$1" in
|
|
51
|
+
--config)
|
|
52
|
+
if [[ -n "$2" ]]; then
|
|
53
|
+
yaml_config="$2"
|
|
54
|
+
shift 2
|
|
55
|
+
else
|
|
56
|
+
echo -e "${RED}Error: $1 requires a CONFIG argument${NC}" >&2
|
|
57
|
+
exit 1
|
|
58
|
+
fi
|
|
59
|
+
;;
|
|
60
|
+
*)
|
|
61
|
+
other_args="$other_args $1"
|
|
62
|
+
shift
|
|
63
|
+
;;
|
|
64
|
+
esac
|
|
65
|
+
done
|
|
66
|
+
|
|
67
|
+
# Check if yaml_config is required
|
|
68
|
+
if [[ "$env_type" == "venv" ]] && [ -z "$yaml_config" ]; then
|
|
69
|
+
echo -e "${RED}Error: --config is required for venv environment${NC}" >&2
|
|
70
|
+
exit 1
|
|
71
|
+
fi
|
|
72
|
+
|
|
73
|
+
PYTHON_BINARY="python"
|
|
74
|
+
case "$env_type" in
|
|
75
|
+
"venv")
|
|
76
|
+
if [ -n "$VIRTUAL_ENV" ] && [ "$VIRTUAL_ENV" == "$env_path_or_name" ]; then
|
|
77
|
+
echo -e "${GREEN}Virtual environment already activated${NC}" >&2
|
|
78
|
+
else
|
|
79
|
+
# Activate virtual environment
|
|
80
|
+
if [ ! -d "$env_path_or_name" ]; then
|
|
81
|
+
echo -e "${RED}Error: Virtual environment not found at $env_path_or_name${NC}" >&2
|
|
82
|
+
exit 1
|
|
83
|
+
fi
|
|
84
|
+
|
|
85
|
+
if [ ! -f "$env_path_or_name/bin/activate" ]; then
|
|
86
|
+
echo -e "${RED}Error: Virtual environment activate binary not found at $env_path_or_name/bin/activate" >&2
|
|
87
|
+
exit 1
|
|
88
|
+
fi
|
|
89
|
+
|
|
90
|
+
source "$env_path_or_name/bin/activate"
|
|
91
|
+
fi
|
|
92
|
+
;;
|
|
93
|
+
*)
|
|
94
|
+
# Handle unsupported env_types here
|
|
95
|
+
echo -e "${RED}Error: Unsupported environment type '$env_type'. Only 'venv' is supported.${NC}" >&2
|
|
96
|
+
exit 1
|
|
97
|
+
;;
|
|
98
|
+
esac
|
|
99
|
+
|
|
100
|
+
if [[ "$env_type" == "venv" ]]; then
|
|
101
|
+
set -x
|
|
102
|
+
|
|
103
|
+
if [ -n "$yaml_config" ]; then
|
|
104
|
+
yaml_config_arg="$yaml_config"
|
|
105
|
+
else
|
|
106
|
+
yaml_config_arg=""
|
|
107
|
+
fi
|
|
108
|
+
|
|
109
|
+
llama stack run \
|
|
110
|
+
$yaml_config_arg \
|
|
111
|
+
--port "$port" \
|
|
112
|
+
$other_args
|
|
113
|
+
elif [[ "$env_type" == "container" ]]; then
|
|
114
|
+
echo -e "${RED}Warning: Llama Stack no longer supports running Containers via the 'llama stack run' command.${NC}"
|
|
115
|
+
echo -e "Please refer to the documentation for more information: https://llamastack.github.io/latest/distributions/building_distro.html#llama-stack-build"
|
|
116
|
+
exit 1
|
|
117
|
+
fi
|
|
@@ -0,0 +1,283 @@
|
|
|
1
|
+
# Copyright (c) Meta Platforms, Inc. and affiliates.
|
|
2
|
+
# All rights reserved.
|
|
3
|
+
#
|
|
4
|
+
# This source code is licensed under the terms described in the LICENSE file in
|
|
5
|
+
# the root directory of this source tree.
|
|
6
|
+
|
|
7
|
+
import re
|
|
8
|
+
from abc import abstractmethod
|
|
9
|
+
from enum import StrEnum
|
|
10
|
+
from pathlib import Path
|
|
11
|
+
from typing import Annotated, Literal
|
|
12
|
+
|
|
13
|
+
from pydantic import BaseModel, Field, field_validator
|
|
14
|
+
|
|
15
|
+
|
|
16
|
+
class StorageBackendType(StrEnum):
|
|
17
|
+
KV_REDIS = "kv_redis"
|
|
18
|
+
KV_SQLITE = "kv_sqlite"
|
|
19
|
+
KV_POSTGRES = "kv_postgres"
|
|
20
|
+
KV_MONGODB = "kv_mongodb"
|
|
21
|
+
SQL_SQLITE = "sql_sqlite"
|
|
22
|
+
SQL_POSTGRES = "sql_postgres"
|
|
23
|
+
|
|
24
|
+
|
|
25
|
+
class CommonConfig(BaseModel):
|
|
26
|
+
namespace: str | None = Field(
|
|
27
|
+
default=None,
|
|
28
|
+
description="All keys will be prefixed with this namespace",
|
|
29
|
+
)
|
|
30
|
+
|
|
31
|
+
|
|
32
|
+
class RedisKVStoreConfig(CommonConfig):
|
|
33
|
+
type: Literal[StorageBackendType.KV_REDIS] = StorageBackendType.KV_REDIS
|
|
34
|
+
host: str = "localhost"
|
|
35
|
+
port: int = 6379
|
|
36
|
+
|
|
37
|
+
@property
|
|
38
|
+
def url(self) -> str:
|
|
39
|
+
return f"redis://{self.host}:{self.port}"
|
|
40
|
+
|
|
41
|
+
@classmethod
|
|
42
|
+
def pip_packages(cls) -> list[str]:
|
|
43
|
+
return ["redis"]
|
|
44
|
+
|
|
45
|
+
@classmethod
|
|
46
|
+
def sample_run_config(cls):
|
|
47
|
+
return {
|
|
48
|
+
"type": StorageBackendType.KV_REDIS.value,
|
|
49
|
+
"host": "${env.REDIS_HOST:=localhost}",
|
|
50
|
+
"port": "${env.REDIS_PORT:=6379}",
|
|
51
|
+
}
|
|
52
|
+
|
|
53
|
+
|
|
54
|
+
class SqliteKVStoreConfig(CommonConfig):
|
|
55
|
+
type: Literal[StorageBackendType.KV_SQLITE] = StorageBackendType.KV_SQLITE
|
|
56
|
+
db_path: str = Field(
|
|
57
|
+
description="File path for the sqlite database",
|
|
58
|
+
)
|
|
59
|
+
|
|
60
|
+
@classmethod
|
|
61
|
+
def pip_packages(cls) -> list[str]:
|
|
62
|
+
return ["aiosqlite"]
|
|
63
|
+
|
|
64
|
+
@classmethod
|
|
65
|
+
def sample_run_config(cls, __distro_dir__: str, db_name: str = "kvstore.db"):
|
|
66
|
+
return {
|
|
67
|
+
"type": StorageBackendType.KV_SQLITE.value,
|
|
68
|
+
"db_path": "${env.SQLITE_STORE_DIR:=" + __distro_dir__ + "}/" + db_name,
|
|
69
|
+
}
|
|
70
|
+
|
|
71
|
+
|
|
72
|
+
class PostgresKVStoreConfig(CommonConfig):
|
|
73
|
+
type: Literal[StorageBackendType.KV_POSTGRES] = StorageBackendType.KV_POSTGRES
|
|
74
|
+
host: str = "localhost"
|
|
75
|
+
port: int | str = 5432
|
|
76
|
+
db: str = "llamastack"
|
|
77
|
+
user: str
|
|
78
|
+
password: str | None = None
|
|
79
|
+
ssl_mode: str | None = None
|
|
80
|
+
ca_cert_path: str | None = None
|
|
81
|
+
table_name: str = "llamastack_kvstore"
|
|
82
|
+
|
|
83
|
+
@classmethod
|
|
84
|
+
def sample_run_config(cls, table_name: str = "llamastack_kvstore", **kwargs):
|
|
85
|
+
return {
|
|
86
|
+
"type": StorageBackendType.KV_POSTGRES.value,
|
|
87
|
+
"host": "${env.POSTGRES_HOST:=localhost}",
|
|
88
|
+
"port": "${env.POSTGRES_PORT:=5432}",
|
|
89
|
+
"db": "${env.POSTGRES_DB:=llamastack}",
|
|
90
|
+
"user": "${env.POSTGRES_USER:=llamastack}",
|
|
91
|
+
"password": "${env.POSTGRES_PASSWORD:=llamastack}",
|
|
92
|
+
"table_name": "${env.POSTGRES_TABLE_NAME:=" + table_name + "}",
|
|
93
|
+
}
|
|
94
|
+
|
|
95
|
+
@classmethod
|
|
96
|
+
@field_validator("table_name")
|
|
97
|
+
def validate_table_name(cls, v: str) -> str:
|
|
98
|
+
# PostgreSQL identifiers rules:
|
|
99
|
+
# - Must start with a letter or underscore
|
|
100
|
+
# - Can contain letters, numbers, and underscores
|
|
101
|
+
# - Maximum length is 63 bytes
|
|
102
|
+
pattern = r"^[a-zA-Z_][a-zA-Z0-9_]*$"
|
|
103
|
+
if not re.match(pattern, v):
|
|
104
|
+
raise ValueError(
|
|
105
|
+
"Invalid table name. Must start with letter or underscore and contain only letters, numbers, and underscores"
|
|
106
|
+
)
|
|
107
|
+
if len(v) > 63:
|
|
108
|
+
raise ValueError("Table name must be less than 63 characters")
|
|
109
|
+
return v
|
|
110
|
+
|
|
111
|
+
@classmethod
|
|
112
|
+
def pip_packages(cls) -> list[str]:
|
|
113
|
+
return ["psycopg2-binary"]
|
|
114
|
+
|
|
115
|
+
|
|
116
|
+
class MongoDBKVStoreConfig(CommonConfig):
|
|
117
|
+
type: Literal[StorageBackendType.KV_MONGODB] = StorageBackendType.KV_MONGODB
|
|
118
|
+
host: str = "localhost"
|
|
119
|
+
port: int = 27017
|
|
120
|
+
db: str = "llamastack"
|
|
121
|
+
user: str | None = None
|
|
122
|
+
password: str | None = None
|
|
123
|
+
collection_name: str = "llamastack_kvstore"
|
|
124
|
+
|
|
125
|
+
@classmethod
|
|
126
|
+
def pip_packages(cls) -> list[str]:
|
|
127
|
+
return ["pymongo"]
|
|
128
|
+
|
|
129
|
+
@classmethod
|
|
130
|
+
def sample_run_config(cls, collection_name: str = "llamastack_kvstore"):
|
|
131
|
+
return {
|
|
132
|
+
"type": StorageBackendType.KV_MONGODB.value,
|
|
133
|
+
"host": "${env.MONGODB_HOST:=localhost}",
|
|
134
|
+
"port": "${env.MONGODB_PORT:=5432}",
|
|
135
|
+
"db": "${env.MONGODB_DB}",
|
|
136
|
+
"user": "${env.MONGODB_USER}",
|
|
137
|
+
"password": "${env.MONGODB_PASSWORD}",
|
|
138
|
+
"collection_name": "${env.MONGODB_COLLECTION_NAME:=" + collection_name + "}",
|
|
139
|
+
}
|
|
140
|
+
|
|
141
|
+
|
|
142
|
+
class SqlAlchemySqlStoreConfig(BaseModel):
|
|
143
|
+
@property
|
|
144
|
+
@abstractmethod
|
|
145
|
+
def engine_str(self) -> str: ...
|
|
146
|
+
|
|
147
|
+
# TODO: move this when we have a better way to specify dependencies with internal APIs
|
|
148
|
+
@classmethod
|
|
149
|
+
def pip_packages(cls) -> list[str]:
|
|
150
|
+
return ["sqlalchemy[asyncio]"]
|
|
151
|
+
|
|
152
|
+
|
|
153
|
+
class SqliteSqlStoreConfig(SqlAlchemySqlStoreConfig):
|
|
154
|
+
type: Literal[StorageBackendType.SQL_SQLITE] = StorageBackendType.SQL_SQLITE
|
|
155
|
+
db_path: str = Field(
|
|
156
|
+
description="Database path, e.g. ~/.llama/distributions/ollama/sqlstore.db",
|
|
157
|
+
)
|
|
158
|
+
|
|
159
|
+
@property
|
|
160
|
+
def engine_str(self) -> str:
|
|
161
|
+
return "sqlite+aiosqlite:///" + Path(self.db_path).expanduser().as_posix()
|
|
162
|
+
|
|
163
|
+
@classmethod
|
|
164
|
+
def sample_run_config(cls, __distro_dir__: str, db_name: str = "sqlstore.db"):
|
|
165
|
+
return {
|
|
166
|
+
"type": StorageBackendType.SQL_SQLITE.value,
|
|
167
|
+
"db_path": "${env.SQLITE_STORE_DIR:=" + __distro_dir__ + "}/" + db_name,
|
|
168
|
+
}
|
|
169
|
+
|
|
170
|
+
@classmethod
|
|
171
|
+
def pip_packages(cls) -> list[str]:
|
|
172
|
+
return super().pip_packages() + ["aiosqlite"]
|
|
173
|
+
|
|
174
|
+
|
|
175
|
+
class PostgresSqlStoreConfig(SqlAlchemySqlStoreConfig):
|
|
176
|
+
type: Literal[StorageBackendType.SQL_POSTGRES] = StorageBackendType.SQL_POSTGRES
|
|
177
|
+
host: str = "localhost"
|
|
178
|
+
port: int | str = 5432
|
|
179
|
+
db: str = "llamastack"
|
|
180
|
+
user: str
|
|
181
|
+
password: str | None = None
|
|
182
|
+
|
|
183
|
+
@property
|
|
184
|
+
def engine_str(self) -> str:
|
|
185
|
+
return f"postgresql+asyncpg://{self.user}:{self.password}@{self.host}:{self.port}/{self.db}"
|
|
186
|
+
|
|
187
|
+
@classmethod
|
|
188
|
+
def pip_packages(cls) -> list[str]:
|
|
189
|
+
return super().pip_packages() + ["asyncpg"]
|
|
190
|
+
|
|
191
|
+
@classmethod
|
|
192
|
+
def sample_run_config(cls, **kwargs):
|
|
193
|
+
return {
|
|
194
|
+
"type": StorageBackendType.SQL_POSTGRES.value,
|
|
195
|
+
"host": "${env.POSTGRES_HOST:=localhost}",
|
|
196
|
+
"port": "${env.POSTGRES_PORT:=5432}",
|
|
197
|
+
"db": "${env.POSTGRES_DB:=llamastack}",
|
|
198
|
+
"user": "${env.POSTGRES_USER:=llamastack}",
|
|
199
|
+
"password": "${env.POSTGRES_PASSWORD:=llamastack}",
|
|
200
|
+
}
|
|
201
|
+
|
|
202
|
+
|
|
203
|
+
# reference = (backend_name, table_name)
|
|
204
|
+
class SqlStoreReference(BaseModel):
|
|
205
|
+
"""A reference to a 'SQL-like' persistent store. A table name must be provided."""
|
|
206
|
+
|
|
207
|
+
table_name: str = Field(
|
|
208
|
+
description="Name of the table to use for the SqlStore",
|
|
209
|
+
)
|
|
210
|
+
|
|
211
|
+
backend: str = Field(
|
|
212
|
+
description="Name of backend from storage.backends",
|
|
213
|
+
)
|
|
214
|
+
|
|
215
|
+
|
|
216
|
+
# reference = (backend_name, namespace)
|
|
217
|
+
class KVStoreReference(BaseModel):
|
|
218
|
+
"""A reference to a 'key-value' persistent store. A namespace must be provided."""
|
|
219
|
+
|
|
220
|
+
namespace: str = Field(
|
|
221
|
+
description="Key prefix for KVStore backends",
|
|
222
|
+
)
|
|
223
|
+
|
|
224
|
+
backend: str = Field(
|
|
225
|
+
description="Name of backend from storage.backends",
|
|
226
|
+
)
|
|
227
|
+
|
|
228
|
+
|
|
229
|
+
StorageBackendConfig = Annotated[
|
|
230
|
+
RedisKVStoreConfig
|
|
231
|
+
| SqliteKVStoreConfig
|
|
232
|
+
| PostgresKVStoreConfig
|
|
233
|
+
| MongoDBKVStoreConfig
|
|
234
|
+
| SqliteSqlStoreConfig
|
|
235
|
+
| PostgresSqlStoreConfig,
|
|
236
|
+
Field(discriminator="type"),
|
|
237
|
+
]
|
|
238
|
+
|
|
239
|
+
|
|
240
|
+
class InferenceStoreReference(SqlStoreReference):
|
|
241
|
+
"""Inference store configuration with queue tuning."""
|
|
242
|
+
|
|
243
|
+
max_write_queue_size: int = Field(
|
|
244
|
+
default=10000,
|
|
245
|
+
description="Max queued writes for inference store",
|
|
246
|
+
)
|
|
247
|
+
num_writers: int = Field(
|
|
248
|
+
default=4,
|
|
249
|
+
description="Number of concurrent background writers",
|
|
250
|
+
)
|
|
251
|
+
|
|
252
|
+
|
|
253
|
+
class ResponsesStoreReference(InferenceStoreReference):
|
|
254
|
+
"""Responses store configuration with queue tuning."""
|
|
255
|
+
|
|
256
|
+
|
|
257
|
+
class ServerStoresConfig(BaseModel):
|
|
258
|
+
metadata: KVStoreReference | None = Field(
|
|
259
|
+
default=None,
|
|
260
|
+
description="Metadata store configuration (uses KV backend)",
|
|
261
|
+
)
|
|
262
|
+
inference: InferenceStoreReference | None = Field(
|
|
263
|
+
default=None,
|
|
264
|
+
description="Inference store configuration (uses SQL backend)",
|
|
265
|
+
)
|
|
266
|
+
conversations: SqlStoreReference | None = Field(
|
|
267
|
+
default=None,
|
|
268
|
+
description="Conversations store configuration (uses SQL backend)",
|
|
269
|
+
)
|
|
270
|
+
responses: ResponsesStoreReference | None = Field(
|
|
271
|
+
default=None,
|
|
272
|
+
description="Responses store configuration (uses SQL backend)",
|
|
273
|
+
)
|
|
274
|
+
|
|
275
|
+
|
|
276
|
+
class StorageConfig(BaseModel):
|
|
277
|
+
backends: dict[str, StorageBackendConfig] = Field(
|
|
278
|
+
description="Named backend configurations (e.g., 'default', 'cache')",
|
|
279
|
+
)
|
|
280
|
+
stores: ServerStoresConfig = Field(
|
|
281
|
+
default_factory=lambda: ServerStoresConfig(),
|
|
282
|
+
description="Named references to storage backends used by the stack core",
|
|
283
|
+
)
|
|
@@ -0,0 +1,199 @@
|
|
|
1
|
+
# Copyright (c) Meta Platforms, Inc. and affiliates.
|
|
2
|
+
# All rights reserved.
|
|
3
|
+
#
|
|
4
|
+
# This source code is licensed under the terms described in the LICENSE file in
|
|
5
|
+
# the root directory of this source tree.
|
|
6
|
+
|
|
7
|
+
import asyncio
|
|
8
|
+
from contextlib import asynccontextmanager
|
|
9
|
+
from typing import Protocol
|
|
10
|
+
|
|
11
|
+
import pydantic
|
|
12
|
+
|
|
13
|
+
from llama_stack.core.datatypes import RoutableObjectWithProvider
|
|
14
|
+
from llama_stack.core.storage.datatypes import KVStoreReference
|
|
15
|
+
from llama_stack.log import get_logger
|
|
16
|
+
from llama_stack.providers.utils.kvstore import KVStore, kvstore_impl
|
|
17
|
+
|
|
18
|
+
logger = get_logger(__name__, category="core::registry")
|
|
19
|
+
|
|
20
|
+
|
|
21
|
+
class DistributionRegistry(Protocol):
|
|
22
|
+
async def get_all(self) -> list[RoutableObjectWithProvider]: ...
|
|
23
|
+
|
|
24
|
+
async def initialize(self) -> None: ...
|
|
25
|
+
|
|
26
|
+
async def get(self, identifier: str) -> RoutableObjectWithProvider | None: ...
|
|
27
|
+
|
|
28
|
+
def get_cached(self, identifier: str) -> RoutableObjectWithProvider | None: ...
|
|
29
|
+
|
|
30
|
+
async def update(self, obj: RoutableObjectWithProvider) -> RoutableObjectWithProvider: ...
|
|
31
|
+
|
|
32
|
+
async def register(self, obj: RoutableObjectWithProvider) -> bool: ...
|
|
33
|
+
|
|
34
|
+
async def delete(self, type: str, identifier: str) -> None: ...
|
|
35
|
+
|
|
36
|
+
|
|
37
|
+
REGISTER_PREFIX = "distributions:registry"
|
|
38
|
+
KEY_VERSION = "v10"
|
|
39
|
+
KEY_FORMAT = f"{REGISTER_PREFIX}:{KEY_VERSION}::" + "{type}:{identifier}"
|
|
40
|
+
|
|
41
|
+
|
|
42
|
+
def _get_registry_key_range() -> tuple[str, str]:
|
|
43
|
+
"""Returns the start and end keys for the registry range query."""
|
|
44
|
+
start_key = f"{REGISTER_PREFIX}:{KEY_VERSION}"
|
|
45
|
+
return start_key, f"{start_key}\xff"
|
|
46
|
+
|
|
47
|
+
|
|
48
|
+
def _parse_registry_values(values: list[str]) -> list[RoutableObjectWithProvider]:
|
|
49
|
+
"""Utility function to parse registry values into RoutableObjectWithProvider objects."""
|
|
50
|
+
all_objects = []
|
|
51
|
+
for value in values:
|
|
52
|
+
try:
|
|
53
|
+
obj = pydantic.TypeAdapter(RoutableObjectWithProvider).validate_json(value)
|
|
54
|
+
all_objects.append(obj)
|
|
55
|
+
except pydantic.ValidationError as e:
|
|
56
|
+
logger.error(f"Error parsing registry value, raw value: {value}. Error: {e}")
|
|
57
|
+
continue
|
|
58
|
+
|
|
59
|
+
return all_objects
|
|
60
|
+
|
|
61
|
+
|
|
62
|
+
class DiskDistributionRegistry(DistributionRegistry):
|
|
63
|
+
def __init__(self, kvstore: KVStore):
|
|
64
|
+
self.kvstore = kvstore
|
|
65
|
+
|
|
66
|
+
async def initialize(self) -> None:
|
|
67
|
+
pass
|
|
68
|
+
|
|
69
|
+
def get_cached(self, type: str, identifier: str) -> RoutableObjectWithProvider | None:
|
|
70
|
+
# Disk registry does not have a cache
|
|
71
|
+
raise NotImplementedError("Disk registry does not have a cache")
|
|
72
|
+
|
|
73
|
+
async def get_all(self) -> list[RoutableObjectWithProvider]:
|
|
74
|
+
start_key, end_key = _get_registry_key_range()
|
|
75
|
+
values = await self.kvstore.values_in_range(start_key, end_key)
|
|
76
|
+
return _parse_registry_values(values)
|
|
77
|
+
|
|
78
|
+
async def get(self, type: str, identifier: str) -> RoutableObjectWithProvider | None:
|
|
79
|
+
json_str = await self.kvstore.get(KEY_FORMAT.format(type=type, identifier=identifier))
|
|
80
|
+
if not json_str:
|
|
81
|
+
return None
|
|
82
|
+
|
|
83
|
+
try:
|
|
84
|
+
return pydantic.TypeAdapter(RoutableObjectWithProvider).validate_json(json_str)
|
|
85
|
+
except pydantic.ValidationError as e:
|
|
86
|
+
logger.error(f"Error parsing registry value for {type}:{identifier}, raw value: {json_str}. Error: {e}")
|
|
87
|
+
return None
|
|
88
|
+
|
|
89
|
+
async def update(self, obj: RoutableObjectWithProvider) -> None:
|
|
90
|
+
await self.kvstore.set(
|
|
91
|
+
KEY_FORMAT.format(type=obj.type, identifier=obj.identifier),
|
|
92
|
+
obj.model_dump_json(),
|
|
93
|
+
)
|
|
94
|
+
return obj
|
|
95
|
+
|
|
96
|
+
async def register(self, obj: RoutableObjectWithProvider) -> bool:
|
|
97
|
+
existing_obj = await self.get(obj.type, obj.identifier)
|
|
98
|
+
if existing_obj and existing_obj != obj:
|
|
99
|
+
raise ValueError(
|
|
100
|
+
f"Object of type '{obj.type}' and identifier '{obj.identifier}' already exists. "
|
|
101
|
+
"Unregister it first if you want to replace it."
|
|
102
|
+
)
|
|
103
|
+
|
|
104
|
+
await self.kvstore.set(
|
|
105
|
+
KEY_FORMAT.format(type=obj.type, identifier=obj.identifier),
|
|
106
|
+
obj.model_dump_json(),
|
|
107
|
+
)
|
|
108
|
+
return True
|
|
109
|
+
|
|
110
|
+
async def delete(self, type: str, identifier: str) -> None:
|
|
111
|
+
await self.kvstore.delete(KEY_FORMAT.format(type=type, identifier=identifier))
|
|
112
|
+
|
|
113
|
+
|
|
114
|
+
class CachedDiskDistributionRegistry(DiskDistributionRegistry):
|
|
115
|
+
def __init__(self, kvstore: KVStore):
|
|
116
|
+
super().__init__(kvstore)
|
|
117
|
+
self.cache: dict[tuple[str, str], RoutableObjectWithProvider] = {}
|
|
118
|
+
self._initialized = False
|
|
119
|
+
self._initialize_lock = asyncio.Lock()
|
|
120
|
+
self._cache_lock = asyncio.Lock()
|
|
121
|
+
|
|
122
|
+
@asynccontextmanager
|
|
123
|
+
async def _locked_cache(self):
|
|
124
|
+
"""Context manager for safely accessing the cache with a lock."""
|
|
125
|
+
async with self._cache_lock:
|
|
126
|
+
yield self.cache
|
|
127
|
+
|
|
128
|
+
async def _ensure_initialized(self):
|
|
129
|
+
"""Ensures the registry is initialized before operations."""
|
|
130
|
+
if self._initialized:
|
|
131
|
+
return
|
|
132
|
+
|
|
133
|
+
async with self._initialize_lock:
|
|
134
|
+
if self._initialized:
|
|
135
|
+
return
|
|
136
|
+
|
|
137
|
+
start_key, end_key = _get_registry_key_range()
|
|
138
|
+
values = await self.kvstore.values_in_range(start_key, end_key)
|
|
139
|
+
objects = _parse_registry_values(values)
|
|
140
|
+
|
|
141
|
+
async with self._locked_cache() as cache:
|
|
142
|
+
for obj in objects:
|
|
143
|
+
cache_key = (obj.type, obj.identifier)
|
|
144
|
+
cache[cache_key] = obj
|
|
145
|
+
|
|
146
|
+
self._initialized = True
|
|
147
|
+
|
|
148
|
+
async def initialize(self) -> None:
|
|
149
|
+
await self._ensure_initialized()
|
|
150
|
+
|
|
151
|
+
def get_cached(self, type: str, identifier: str) -> RoutableObjectWithProvider | None:
|
|
152
|
+
return self.cache.get((type, identifier), None)
|
|
153
|
+
|
|
154
|
+
async def get_all(self) -> list[RoutableObjectWithProvider]:
|
|
155
|
+
await self._ensure_initialized()
|
|
156
|
+
async with self._locked_cache() as cache:
|
|
157
|
+
return list(cache.values())
|
|
158
|
+
|
|
159
|
+
async def get(self, type: str, identifier: str) -> RoutableObjectWithProvider | None:
|
|
160
|
+
await self._ensure_initialized()
|
|
161
|
+
cache_key = (type, identifier)
|
|
162
|
+
|
|
163
|
+
async with self._locked_cache() as cache:
|
|
164
|
+
return cache.get(cache_key, None)
|
|
165
|
+
|
|
166
|
+
async def register(self, obj: RoutableObjectWithProvider) -> bool:
|
|
167
|
+
await self._ensure_initialized()
|
|
168
|
+
success = await super().register(obj)
|
|
169
|
+
|
|
170
|
+
if success:
|
|
171
|
+
cache_key = (obj.type, obj.identifier)
|
|
172
|
+
async with self._locked_cache() as cache:
|
|
173
|
+
cache[cache_key] = obj
|
|
174
|
+
|
|
175
|
+
return success
|
|
176
|
+
|
|
177
|
+
async def update(self, obj: RoutableObjectWithProvider) -> None:
|
|
178
|
+
await super().update(obj)
|
|
179
|
+
cache_key = (obj.type, obj.identifier)
|
|
180
|
+
async with self._locked_cache() as cache:
|
|
181
|
+
cache[cache_key] = obj
|
|
182
|
+
return obj
|
|
183
|
+
|
|
184
|
+
async def delete(self, type: str, identifier: str) -> None:
|
|
185
|
+
await super().delete(type, identifier)
|
|
186
|
+
cache_key = (type, identifier)
|
|
187
|
+
async with self._locked_cache() as cache:
|
|
188
|
+
if cache_key in cache:
|
|
189
|
+
del cache[cache_key]
|
|
190
|
+
|
|
191
|
+
|
|
192
|
+
async def create_dist_registry(
|
|
193
|
+
metadata_store: KVStoreReference, image_name: str
|
|
194
|
+
) -> tuple[CachedDiskDistributionRegistry, KVStore]:
|
|
195
|
+
# instantiate kvstore for storing and retrieving distribution metadata
|
|
196
|
+
dist_kvstore = await kvstore_impl(metadata_store)
|
|
197
|
+
dist_registry = CachedDiskDistributionRegistry(dist_kvstore)
|
|
198
|
+
await dist_registry.initialize()
|
|
199
|
+
return dist_registry, dist_kvstore
|
|
@@ -0,0 +1,49 @@
|
|
|
1
|
+
# Copyright (c) Meta Platforms, Inc. and affiliates.
|
|
2
|
+
# All rights reserved.
|
|
3
|
+
#
|
|
4
|
+
# This source code is licensed under the terms described in the LICENSE file in
|
|
5
|
+
# the root directory of this source tree.
|
|
6
|
+
|
|
7
|
+
import os
|
|
8
|
+
from contextvars import ContextVar
|
|
9
|
+
|
|
10
|
+
from llama_stack.core.request_headers import PROVIDER_DATA_VAR
|
|
11
|
+
|
|
12
|
+
TEST_CONTEXT: ContextVar[str | None] = ContextVar("llama_stack_test_context", default=None)
|
|
13
|
+
|
|
14
|
+
|
|
15
|
+
def get_test_context() -> str | None:
|
|
16
|
+
return TEST_CONTEXT.get()
|
|
17
|
+
|
|
18
|
+
|
|
19
|
+
def set_test_context(value: str | None):
|
|
20
|
+
return TEST_CONTEXT.set(value)
|
|
21
|
+
|
|
22
|
+
|
|
23
|
+
def reset_test_context(token) -> None:
|
|
24
|
+
TEST_CONTEXT.reset(token)
|
|
25
|
+
|
|
26
|
+
|
|
27
|
+
def sync_test_context_from_provider_data():
|
|
28
|
+
"""Sync test context from provider data when running in server test mode."""
|
|
29
|
+
if "LLAMA_STACK_TEST_INFERENCE_MODE" not in os.environ:
|
|
30
|
+
return None
|
|
31
|
+
|
|
32
|
+
stack_config_type = os.environ.get("LLAMA_STACK_TEST_STACK_CONFIG_TYPE", "library_client")
|
|
33
|
+
if stack_config_type != "server":
|
|
34
|
+
return None
|
|
35
|
+
|
|
36
|
+
try:
|
|
37
|
+
provider_data = PROVIDER_DATA_VAR.get()
|
|
38
|
+
except LookupError:
|
|
39
|
+
provider_data = None
|
|
40
|
+
|
|
41
|
+
if provider_data and "__test_id" in provider_data:
|
|
42
|
+
return TEST_CONTEXT.set(provider_data["__test_id"])
|
|
43
|
+
|
|
44
|
+
return None
|
|
45
|
+
|
|
46
|
+
|
|
47
|
+
def is_debug_mode() -> bool:
|
|
48
|
+
"""Check if test recording debug mode is enabled via LLAMA_STACK_TEST_DEBUG env var."""
|
|
49
|
+
return os.environ.get("LLAMA_STACK_TEST_DEBUG", "").lower() in ("1", "true", "yes")
|