[UNIT] llama-index framework - release 3/3/25 #22
Conversation
|
Looks nice. I don't see anything on:
|
Yes, I've updated the ToC outside of here in a Google Docs, will update based on the feedback of |
burtenshaw
changed the title
|
@davidberenstein1957 We'll release this unit in 3 weeks. |
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
…e LangChain references
…and Agent creation guide
…xamples and restructuring
…components, agents, and workflows
…ti-agent system explanations
There was a problem hiding this comment.
Good work, thanks to that I didn't have a lot to update but some quick questions:
-
Do you provide notebooks for users to rapidly try llama-index? It's what we did for smolagents and it really helps usage (because low friction since nothing to install on your computer).
-
If yes, we should add the notebooks to agentcourse/notebooks in HF + put links in the sections like we did with smolagents (check smolagents sections to see an example).
Also there was some typos, display problems don't hesitate to check the final look here: https://moon-ci-docs.huggingface.co/learn/agents-course/pr_22/unit2/llama-index/components
|
@simoninithomas yea we have tons of notebooks -- although looking at the smolagents notebooks, it looks pretty tailored to the course. I can copy those but just migrate the code/concepts to llama-index? Or any other ideas/thoughts on that? |
- Enhance components and tools notebooks with more detailed code examples - Update installation instructions and code formatting - Add more descriptive comments and improve code readability - Minor updates to MDX documentation to reflect notebook changes
…examples - Add dependencies installation section - Implement basic agent initialization with multiplication tool - Demonstrate context preservation in agent interactions - Create RAG agent using QueryEngineTool - Develop multi-agent system with calculator and query agents - Update notebook metadata and kernel configuration
- Migrate to Hugging Face Inference API for LLM - Add comprehensive workflow examples with multi-step and branching logic - Implement workflow drawing functionality using pyvis - Update notebook with state management and multi-agent workflow demonstrations - Enhance code with more detailed comments and type hinting
…ebook - Delete redundant flow visualization HTML files (flow.html and workflow_all_flows.html) - Remove empty code cell from workflows notebook - Simplify workflow example in MDX documentation
There was a problem hiding this comment.
super uber nit, but the spacing in some of the code (like in the query engine section) is triggering my OCD 😆 (I'd fix it myself, but I don't think I have permissions to push to this branch)
There was a problem hiding this comment.
I might actually set verbose=False on the AgentWorkflow, the output is a little hard to read when we are also streaming the response
There was a problem hiding this comment.
On the Loops and Branches section, the workflow doesn't actually loop
Needed to update the type annotations on step_one()
class MultiStepWorkflow(Workflow):
@step
async def step_one(self, ev: StartEvent | LoopEvent) -> ProcessingEvent | LoopEvent:
if random.randint(0, 1) == 0:
print("Bad thing happened")
return LoopEvent(loop_output="Back to step one.")
else:
print("Good thing happened")
return ProcessingEvent(intermediate_result="First step complete.")
@step
async def step_two(self, ev: ProcessingEvent) -> StopEvent:
# Use the intermediate result
final_result = f"Finished processing: {ev.intermediate_result}"
return StopEvent(result=final_result)
Co-authored-by: Logan <[email protected]>
- Update function calls and method signatures in agents, components, tools, and workflows notebooks - Improve parameter naming and add explicit keyword arguments - Simplify code structure and remove unnecessary verbose settings - Update type hints and function annotations - Enhance code readability and consistency across notebooks
- Add Tip component with Google Colab notebook links for agents, components, tools, and workflows MDX files - Provide direct links to corresponding Jupyter notebooks for easy code reference - Enhance documentation with practical code exploration resources
|
In tools.mdx, there are some discrepancies between the notebook and the theory:
|
No description provided.