Piyush Jain
Piyush Jain
## Problem The current magics implementation is using an artificial list of chat message history to keep track of exchanges between 2 or more cells when `openai-chat` model is used....
Support for Bedrock Agents. https://docs.aws.amazon.com/bedrock/latest/userguide/agents.html
## Problem Current standard unit and integration tests are not running for all providers, only Anthropic, Meta and Mistral and skipping some tests as well (4/19 are skipped). ## Solution...
Add standard integration test suite for Amazon, AI21, Cohere, and Meta models for the ChatBedrock and ChatBedrockConverse classes. For example, here is the standard test for Anthropic models. ```python class...
3/63 unit tests are marked `xfail` for ChatBedrock and ChatBedrockConverse. ### Test Log ```shell tests/unit_tests/chat_models/test_bedrock_converse.py::TestBedrockStandard::test_init_streaming FAILED tests/unit_tests/test_standard.py::TestBedrockStandard::test_standard_params FAILED tests/unit_tests/test_standard.py::TestBedrockAsConverseStandard::test_standard_params FAILED ==================================================================== FAILURES ===================================================================== _____________________________________________________ TestBedrockStandard.test_init_streaming _____________________________________________________ self = @pytest.mark.xfail(reason="Doesn't support...
Add `stop` attribute to ChatBedrock and ChatBedrockConverse classes to fix the standard "test_stop_sequence" integration tests. ```shell tests/integration_tests/chat_models/test_standard.py::TestBedrockStandard::test_stop_sequence FAILED tests/integration_tests/chat_models/test_standard.py::TestBedrockUseConverseStandard::test_stop_sequence FAILED ```
The standard integration test "test_tool_message_histories_list_content" is failing for ChatBedrockConverse for Mistral models. ### Error log ```shell ======================================================================= FAILURES ======================================================================= _________________________________________ TestBedrockMistralStandard.test_tool_message_histories_list_content __________________________________________ self = model = ChatBedrockConverse(disable_streaming='tool_calling', client=, m...equences=[], temperature=0.0,...
## Problem Bedrock supports several providers (AI21, Anthropic, Amazon, Mistral, Meta etc.), each of which has a slightly different request/response structure. The current ChatBedrock class currently accommodates these providers by...
Some of the tests marked `xfail` are passing in my local testing. Enabling them, so these run during the release. ```shell tests/integration_tests/chat_models/test_standard.py::TestBedrockStandard::test_tool_message_histories_string_content PASSED tests/integration_tests/chat_models/test_standard.py::TestBedrockStandard::test_tool_message_histories_list_content PASSED tests/integration_tests/chat_models/test_standard.py::TestBedrockStandard::test_structured_few_shot_examples PASSED ```