0.19.3

What's Changed

• Default Ollama to use llama3.2 by @WToa in #889
• Apply workaround to Chroma query by @justi in #890
• [SECURITY] Bump rails-html-sanitizer from 1.6.0 to 1.6.1 by @dependabot in #893
• Langchain::Tool::Database#describe_table with COMMENT by @Bahanix in #891
• [SECURITY] Bump actionpack from 7.1.3.3 to 7.1.5.1 by @dependabot in #895
• [SECURITY] Update dotenv-rails requirement from ~> 2.7.6 to ~> 3.1.6 by @dependabot in #899
• Empty text content should not be set when content is nil when using AnthropicMessage by @andreibondarev in #900
• 0.19.3 by @andreibondarev in #905

New Contributors

• @justi made their first contribution in #890
• @Bahanix made their first contribution in #891

Full Changelog: 0.19.2...0.19.3

0.19.2

What's Changed

• Add assistant.tool_execution_callback by @andreibondarev in #884
• 0.19.2 by @andreibondarev in #885
• Fix wrong published version by @andreibondarev in #888

Full Changelog: 0.19.1...0.19.2

0.19.1

What's Changed

• replace include with extend in instructions on how to create custom tools by @jeduardo824 in #857
• Update README.md by @palladius in #862
• [SECURITY] Bump rexml from 3.3.6 to 3.3.9 by @dependabot in #863
• Anthropic accepts messages with images by @andreibondarev in #858
• Improve passing max tokens to anthropic by @andreibondarev in #864
• Langchain::Assistant works with AWS Bedrock-hosted Anthropic models by @andreibondarev in #849
• Enhancement: Refactor for to_hash function in different messages class to be split by role by @rishirajkumar97 in #867
• Anthropic tool use is always accompanied with text content. We should be sending it back as well. by @andreibondarev in #869
• 847 add ability to send images to the assistant when using ollama by @sergiobayona in #870
• Allow passing in options hash to Ollama by @chalmagean in #871
• 0.19.1 by @andreibondarev in #879

New Contributors

• @jeduardo824 made their first contribution in #857
• @rishirajkumar97 made their first contribution in #867
• @sergiobayona made their first contribution in #870

Full Changelog: 0.19.0...0.19.1

0.19.0

What's Changed

• Anthropic Tools: added default input_schema if no params by @dghirardo in #837
• Fix issue with clearing instructions by @qarol in #836
• Allow setting safety_setting: [] in default_options for Langchain::LLM::GoogleGemini and Langchain::LLM::GoogleVertexAI constructors by @andreibondarev in #838
• Enhance code comments by @andreibondarev in #846
• Fix MistralAIMessage to handle "Tool" Output by @dkaplan88 in #850
• Shorter model name params by @bborn in #840
• 0.19.0 by @andreibondarev in #854

New Contributors

• @qarol made their first contribution in #836
• @dkaplan88 made their first contribution in #850

Full Changelog: 0.18.0...0.19.0

0.18.0

What's Changed

• Bump anthropic gem by @andreibondarev in #819
• Add streaming support for Anthropic by @chalmagean in #809
• Resolve the issue of the AWS Bedrock chat model always returning nil by @khaile in #801
• Fix zeitwerk error when eager_load is turned on by @andreibondarev in #822
• Refactor assistant determine tool role method by @andreibondarev in #823
• Introduce support_system_message? method by @andreibondarev in #824
• Introduce Langchain::Assistant#parallel_tool_calls options whether to allow the LLM to make multiple parallel tool calls by @andreibondarev in #827
• 0.18.0 by @andreibondarev in #828

Full Changelog: 0.17.1...0.18.0

0.17.1

What's Changed

• Move adapter-related classes to separate files by @khaile in #807
• Fix Langchain::Tool::Database#describe_table method by @andreibondarev in #814
• 0.17.1 by @andreibondarev in #815

Full Changelog: 0.17.0...0.17.1

0.17.0

What's Changed

• Bump webrick from 1.8.1 to 1.8.2 by @dependabot in #797
• Langchain::Assistant when using OpenAI accepts a message with image_url by @andreibondarev in #799
• Langchain::Assistant when using MistralAI accepts a message with image_url by @andreibondarev in #803
• Vectorsearch providers should utilize the global Langchain.logger by @andreibondarev in #804
• Remove Langchain::LLM::GooglePalm by @andreibondarev in #805
• 0.17.0 by @andreibondarev in #806

Full Changelog: 0.16.1...0.17.0

0.16.1

What's Changed

• Add streaming support by @chalmagean in #760
• Deprecate Langchain::LLM::GooglePalm by @andreibondarev in #789
• Allow explicitly setting client object on any instance of Langchain::LLM::Base subclass by @andreibondarev in #788
• Allow setting response_object: {} parameter when initializing supported Langchain::LLM::* classes by @andreibondarev in #766
• Updated README.md by @andreibondarev in #793
• Improved logging by @dferrazm in #796
• 0.16.1 by @dferrazm in #798

Full Changelog: 0.16.0...0.16.1

0.16.0

What's Changed

• Remove Langchain::Thread class by @andreibondarev in #783
• Support cohere provider for Langchain::LLM::AwsBedrock#embed by @khaile in #782
• Output meaning errors when GoogleGemini or GoogleVertexAI fail by @andreibondarev in #784
• Patch up ContextualLogger by @andreibondarev in #785
• 0.16.0 by @andreibondarev in #787

New Contributors

• @khaile made their first contribution in #782

Full Changelog: 0.15.6...0.16.0

0.15.6

What's Changed

• Add Meta models support for AWS Bedrock LLM by @Fodoj in #764
• Update azure_spec.rb by @scottwater in #778
• Assistant fixes by @andreibondarev in #780
• 0.15.6 by @andreibondarev in #781

New Contributors

• @Fodoj made their first contribution in #764
• @scottwater made their first contribution in #778

Full Changelog: 0.15.5...0.15.6