[RFC] v1.3.0 Roadmap

[DefTruth]

Path to v1.3.0

• cache-dit-generate command line tool @DefTruth feat: add cache-dit-generate cli tool #752
• Optimize VAE Parallel comm, use batched isend/irecv @DefTruth chore: use batched isend/irecv for vae-p #757 feat: tile batched p2p comm for vae-p #758 [2/N] reduce comm overhead for vae-p  #763 reduce comm overhead for vae-p #762
• Hybrid CP(USP) + TP, e.g, SP2 + TP2 @DefTruth feat: support hybrid CP/SP + TP #765
• Support USP (hybrid ulysses and ring attention) @DefTruth feat: support ring attn p2p comm #754 feat: support USP -> Ulysses + Ring #755
• New models support: GLM-Image, FLUX.2-Klein, Helios, FireRed-Image-Edit, and more. @DefTruth feat: support tensor parallel for glm-image #794 feat: support cache for glm-image #787 feat: support 🚀flux2-klein series #717 feat: support 🔥FireRed-Image-Edit-1.0 #797 feat: support cache for Helios-14B #834 feat: support FireRed-Image-Edit-1.1 #854
• Support pass a quantize_config to enable_cache API @DefTruth feat: support load quantize config yaml #847
• FP8 Blockwise dynamic quantization support @DefTruth feat: support blockwise fp8 #822
• AMD GPU support feat: Add AMD GPU support #841