Add unified provider interface with string-based initialization

[jxnl]

Summary

• Adds a unified string-based provider interface with from_provider function for simplified initialization of LLM clients
• Automatically selects appropriate default modes for each provider
• Comprehensive error handling with helpful error messages
• Full test coverage and documentation including a blog post

TODOs before merging

• [ ] Add additional test cases for the remaining providers beyond OpenAI
• [ ] Consider adding the Provider enum to the auto_client imports for consistency
• [ ] Add integration examples showing the new string-based initialization
• [ ] Update all provider documentation to mention the new initialization method
• [ ] Add API documentation for from_provider in the API reference

Test plan

• Verify string parsing works correctly for all provider formats
• Ensure default modes are correctly applied
• Test error handling for missing dependencies and invalid provider strings
• Check that async/sync client creation works correctly
• Verify proper parameter passing to provider client functions

🤖 Generated with Claude Code

---

[!IMPORTANT] Introduces a unified provider interface with string-based initialization for LLM clients, including error handling and documentation updates.

• Behavior:
  • Adds from_provider function in auto_client.py for string-based initialization of LLM clients.
  • Supports synchronous and asynchronous clients with async_client parameter.
  • Automatically selects default modes for each provider.
  • Handles errors for missing dependencies and invalid provider strings.
• Documentation:
  • Updates README.md with provider initialization examples.
  • Adds blog post string-based-init.md detailing the new interface.
  • Updates index.md in docs/integrations to include new initialization method.
• Tests:
  • Adds test_auto_client.py to test provider parsing and error handling.
  • Updates test_simple_type_fix.py and test_fizzbuzz_fix.py for Python version compatibility.
• Misc:
  • Adds run.py example in examples/automodel demonstrating the new interface.
  • Updates pyproject.toml version to 1.8.0.

^{This description was created by for d7d0d7a4d382b1936f4cc2566271f891e037e099. You can customize this summary. It will automatically update as commits are pushed.}

[cloudflare-workers-and-pages[bot]]

Deploying instructor-py with    Cloudflare Pages

Latest commit:	d7d0d7a
Status:	✅  Deploy successful!
Preview URL:	https://c0300b30.instructor-py.pages.dev
Branch Preview URL:	https://automodel.instructor-py.pages.dev

View logs

[ivanleomk]

Tests are passing here in CI

tests/test_auto_client.py ..........................                                             [100%]

[jxnl]

Only do literals if it's also unions with the default string.

We don't want to make people upgrade instructor just to use a new model.

[jxnl]

Let's publish first and then fix docs afterwards.

This is blocking some stuff I want to do in instructor-classify