[Release Plan] v0.0.1

[drewjin]

Version

v0.0.1

Milestone

No response

Scope

• Full support for Block Diffusion and D2F inference strategies
• Expand test coverage and refactor the examples/ directory for better usability
• Establish standardized benchmark suites for performance evaluation
• Implement high-level Strategy Abstraction to minimize development overhead
• Deliver comprehensive developer tutorials and refined technical documentation
• Optimize inference engine to ensure production-ready performance and throughput
• Refactor codebase to improve readability and maintainability

Tasks

• Features
• Fixes
• Docs
• API/Breaking changes
• Benchmarks
• Release notes

Readiness

• All planned issues closed or deferred
• Docs updated
• CI green; artifacts verified
• Release notes drafted

Notes

No response

[drewjin]

Please refer to PRs and follow the developing progress:

• Enhance Strategy: [Feat] Enhance Decoding Strategies for Easier Development and More Efficient Inference #18
• Fast-dLLM-v2: [Feat] Optimize D2F Decoding Strategy to Support CUDA Graph and More Efficient Inference #17
• D2F Optimized: [Feat] Fast-dLLM-v2 Decoding Strategy Support #16