PR #21374: Multimodal synthesis

• Repository: run-llama/llama_index
• Author: bschloss
• State: open | merged: False
• Link: https://github.com/run-llama/llama_index/pull/21374

Description (problem / solution / changelog)

Description

Please include a summary of the change and which issue is fixed. Please also include relevant motivation and context. List any dependencies that are required for this change.

First of two or three PRs to broadly support multimodal synthesis. This PR:

1. Adds a BaseMultimodalSynthesizer class to address existing semantic issues with variable naming and logical issues with conversion of nodes to multimodal content.
2. Creates multimodal prompts for relevant synthesizers
3. Adds multimodal synthesizers for ContextOnly, Generation, NoText, SimpleSummarize (relatively basic synthesizers), as well as Refine and CompactAndRefine.
4. Support for streaming StructuredResponse objects was also added to the Refine synthesizer.
5. Adds lacking test support for response synthesizers

Fixes # 21373

Although the number of lines is large in this PR, the total logical changes are not that great. Fairly repetitive across the basic MultimodalSynthesizers. Since the Refine synthesizer contained more complicated updates, I will follow up with a second PR for the remaining synthesizers so focus can be given to the updates there. Also, many lines were added because there was little to no testing of the synthesizer classes. Some suggestions are made in PR comments on how to reduce total bloat here. Unfortunately though, because of some logical and semantic issues with the BaseSynthesizer class, it seemed like a better idea to make a new Multimodal synthesizer class so as to not introduce breaking changes or overly complicated logic/function signatures in the BaseSynthesizer class.

New Package?

Did I fill in the tool.llamahub section in the pyproject.toml and provide a detailed README.md for my new integration or package?

• Yes
• No

Version Bump?

Did I bump the version in the pyproject.toml file of the package I am updating? (Except for the llama-index-core package)

• Yes
• No

Type of Change

Please delete options that are not relevant.

• New feature (non-breaking change which adds functionality)

How Has This Been Tested?

Your pull-request will likely not be merged unless it is covered by some form of impactful unit testing.

• I added new unit tests to cover this change
• I believe this change is already covered by existing unit tests

Suggested Checklist:

• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• I have added Google Colab support for the newly added notebooks.
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• I ran uv run make format; uv run make lint to appease the lint gods

Changed files

• llama-index-core/llama_index/core/indices/prompt_helper.py (modified, +4/-21)
• llama-index-core/llama_index/core/instrumentation/events/synthesis.py (modified, +32/-0)
• llama-index-core/llama_index/core/llms/mock.py (modified, +94/-18)
• llama-index-core/llama_index/core/postprocessor/llm_rerank.py (modified, +2/-2)
• llama-index-core/llama_index/core/program/llm_program.py (modified, +94/-1)
• llama-index-core/llama_index/core/program/utils.py (modified, +7/-5)
• llama-index-core/llama_index/core/prompts/base.py (modified, +1/-11)
• llama-index-core/llama_index/core/prompts/chat_prompts.py (modified, +108/-1)
• llama-index-core/llama_index/core/prompts/prompt_utils.py (modified, +3/-3)
• llama-index-core/llama_index/core/response_synthesizers/base.py (modified, +203/-2)
• llama-index-core/llama_index/core/response_synthesizers/compact_and_refine.py (modified, +58/-4)
• llama-index-core/llama_index/core/response_synthesizers/context_only.py (modified, +43/-1)
• llama-index-core/llama_index/core/response_synthesizers/generation.py (modified, +176/-2)
• llama-index-core/llama_index/core/response_synthesizers/no_text.py (modified, +31/-1)
• llama-index-core/llama_index/core/response_synthesizers/refine.py (modified, +626/-272)
• llama-index-core/llama_index/core/response_synthesizers/simple_summarize.py (modified, +119/-2)
• llama-index-core/llama_index/core/settings.py (modified, +44/-1)
• llama-index-core/tests/chat_engine/test_condense_plus_context.py (modified, +7/-11)
• llama-index-core/tests/chat_engine/test_context.py (modified, +9/-13)
• llama-index-core/tests/indices/response/test_response_builder.py (modified, +2/-1)
• llama-index-core/tests/response_synthesizers/test_compact_and_refine.py (added, +391/-0)
• llama-index-core/tests/response_synthesizers/test_generate.py (modified, +172/-28)
• llama-index-core/tests/response_synthesizers/test_refine.py (modified, +1199/-119)
• llama-index-core/tests/response_synthesizers/test_simple_summarize.py (added, +220/-0)