Python: #6761 Onnx Connector

[nmoeller]

Motivation and Context

1. Why is this changed required ?
   To enable Onnx Models with Semantic Kernel, there was the issue Python: Add support for local models via ONNX #6761 in the Backlog to add a Onnx Connector
2. What problem does it solve ?
   It solves the problem, that semantic kernel is not yet integrated with Onnx Gen Ai Runtime
3. What scenario does it contribute to?
   The scenario is to use different connector than HF,OpenAI or AzureOpenAI. When User's want to use Onnx they can easliy integrate it now
4. If it fixes an open issue, please link to the issue here.
   Python: Add support for local models via ONNX #6761

Description

The changes made are designed by my own based on other connectors, i tried to stay as close as possible to the structure.
For the integration i installed the mistral python package in the repository.

I added the following Classes :

• OnnxCompletionBase --> Responsible to control the inference
• OnnxTextCompletion --> Inherits from OnnxCompletionBase
  • Support for Text Completion with and without Images
  • Ready for Multimodal Inference
  • Ready for Text Only Inference
  • Supports all Models on onnxruntime-genai
• OnnxChatCompletion -->Inherits from OnnxCompletionBase
  • Support for Text Completion with and without Images
  • The user needs to provide the corresponding chat template to use this class
  • Ready for Multimodal Inference
  • Ready for Text Only Inference
  • Supports all Models on onnxruntime-genai

What is integrated yet :

• OnnxCompletionBase Class
• OnnxChatCompletionBase Class with Dynamic Template Support
• OnnxTextCompletionBase Class
• Sample Multimodal Inference with Phi3-Vision
• Sample of OnnxChatCompletions with Phi3
• Sample of OnnxTextCompletions with Phi3
• Integration Tests
• Unit Tests

Some Notes

Contribution Checklist

• The code builds clean without any errors or warnings
• The PR follows the SK Contribution Guidelines and the pre-submission formatting script raises no violations
• All unit tests pass, and I have added new tests where possible
• I didn't break anyone 😄

[TaoChenOSU]

Regarding our offline conversation on the prompt template, is using a prompt template to parse the chat history to some format an overkill? Prompt template can do much more that substituting arguments. Is it possible to override the _prepare_chat_history_for_request method to get what Onnx wants?

[TaoChenOSU]

Thanks for the contribution! Will approve once the unit tests pass.

[markwallace-microsoft]

Python Test Coverage Report

File	Stmts	Miss	Cover	Missing
semantic_kernel
kernel.py	199	47	76%	148, 159, 163, 313–316, 423, 437–480
semantic_kernel/agents/group_chat
agent_chat.py	124	2	98%	78, 171
agent_group_chat.py	100	2	98%	151, 201
broadcast_queue.py	72	1	99%	35
semantic_kernel/agents/open_ai
assistant_content_generation.py	133	9	93%	96–97, 281, 291–294, 335, 337
open_ai_assistant_base.py	449	8	98%	259, 337–338, 746, 867, 870, 932, 990
semantic_kernel/connectors/ai
chat_completion_client_base.py	116	2	98%	382, 392
completion_usage.py	8	1	88%	17
semantic_kernel/connectors/ai/anthropic/services
anthropic_chat_completion.py	176	5	97%	147, 165, 169, 223, 419
semantic_kernel/connectors/ai/azure_ai_inference/services
azure_ai_inference_chat_completion.py	119	7	94%	120, 146–149, 159, 180, 202
azure_ai_inference_text_embedding.py	41	1	98%	87
semantic_kernel/connectors/ai/embeddings
embedding_generator_base.py	8	1	88%	50
semantic_kernel/connectors/ai/google
shared_utils.py	26	1	96%	56
semantic_kernel/connectors/ai/google/google_ai/services
google_ai_chat_completion.py	119	4	97%	127, 153, 176, 178
google_ai_text_completion.py	63	2	97%	98, 121
utils.py	65	3	95%	140, 160–165
semantic_kernel/connectors/ai/google/vertex_ai/services
utils.py	66	3	95%	141, 161–166
vertex_ai_chat_completion.py	119	4	97%	121, 147, 170, 172
vertex_ai_text_completion.py	62	2	97%	95, 116
semantic_kernel/connectors/ai/hugging_face/services
hf_text_completion.py	60	3	95%	103, 112, 127
hf_text_embedding.py	32	5	84%	79–83
semantic_kernel/connectors/ai/mistral_ai/services
mistral_ai_chat_completion.py	118	7	94%	118–121, 307–310
semantic_kernel/connectors/ai/ollama/services
ollama_chat_completion.py	60	5	92%	95–98, 108, 143
ollama_text_completion.py	57	6	89%	72, 90–93, 103, 131
semantic_kernel/connectors/ai/onnx
utils.py	53	3	94%	50–51, 112
semantic_kernel/connectors/ai/onnx/services
onnx_gen_ai_chat_completion.py	72	7	90%	67–68, 98, 122, 167, 173, 179
onnx_gen_ai_completion_base.py	58	21	64%	59–71, 79–90
onnx_gen_ai_text_completion.py	46	5	89%	54–55, 87, 117, 133
semantic_kernel/connectors/ai/open_ai/prompt_execution_settings
open_ai_prompt_execution_settings.py	94	1	99%	112
semantic_kernel/connectors/ai/open_ai/services
azure_chat_completion.py	107	5	95%	118, 123, 157, 166, 169
azure_text_completion.py	28	2	93%	82, 87
azure_text_embedding.py	30	2	93%	84, 89
open_ai_chat_completion_base.py	127	5	96%	71, 121, 141, 177, 287
open_ai_handler.py	63	3	95%	86, 95–96
open_ai_text_completion_base.py	80	2	98%	56, 161
semantic_kernel/connectors/ai/open_ai/settings
azure_open_ai_settings.py	22	1	95%	99
semantic_kernel/connectors/memory/azure_ai_search
azure_ai_search_collection.py	87	2	98%	150, 152
semantic_kernel/connectors/memory/redis
redis_collection.py	160	2	99%	146, 316
utils.py	45	11	76%	145–146, 164, 166, 173–188
semantic_kernel/connectors/openapi_plugin
openapi_manager.py	58	2	97%	110–111
openapi_parser.py	88	2	98%	71, 128
openapi_runner.py	105	2	98%	181–182
semantic_kernel/connectors/openapi_plugin/models
rest_api_operation.py	129	1	99%	242
semantic_kernel/contents
function_call_content.py	97	1	99%	201
streaming_chat_message_content.py	68	1	99%	210
streaming_content_mixin.py	39	2	95%	37, 64
semantic_kernel/core_plugins/sessions_python_tool
sessions_python_plugin.py	134	8	94%	69, 82–91, 99
sessions_python_settings.py	39	4	90%	84–87
semantic_kernel/data
vector_store_record_collection.py	249	19	92%	410, 470–474, 482–486, 526–529, 536–539
vector_store_record_utils.py	26	2	92%	50, 52
semantic_kernel/functions
kernel_function_decorator.py	98	1	99%	102
kernel_function_from_method.py	96	1	99%	153
kernel_function_from_prompt.py	154	7	95%	165–166, 180, 201, 219, 239, 322
kernel_function_log_messages.py	36	6	83%	37–43
kernel_plugin.py	187	2	99%	472, 475
semantic_kernel/planners
plan.py	234	45	81%	54, 163–165, 197, 214–227, 264, 269, 277–278, 288–291, 308, 313, 329, 332–337, 355, 360, 363, 365, 372, 386–388, 393–397
semantic_kernel/planners/function_calling_stepwise_planner
function_calling_stepwise_planner.py	116	4	97%	145, 189–190, 198
semantic_kernel/planners/sequential_planner
sequential_planner.py	64	6	91%	71, 75, 109, 125, 134–135
sequential_planner_extensions.py	50	9	82%	31–32, 56, 110–124
sequential_planner_parser.py	77	12	84%	66–74, 93, 117–120
semantic_kernel/schema
kernel_json_schema_builder.py	129	9	93%	53, 90, 186, 194, 205, 213, 228, 232–233
semantic_kernel/services
ai_service_client_base.py	22	1	95%	64
semantic_kernel/template_engine/blocks
code_block.py	77	1	99%	119
named_arg_block.py	43	1	98%	98
semantic_kernel/utils/authentication
entra_id_authentication.py	15	2	87%	26, 38
semantic_kernel/utils/telemetry
user_agent.py	16	2	88%	18–19
semantic_kernel/utils/telemetry/model_diagnostics
decorators.py	171	4	98%	372–375
TOTAL	11692	360	97%

Python Unit Test Overview

Tests	Skipped	Failures	Errors	Time
2560	4 💤	0 ❌	0 🔥	1m 5s ⏱️