Haopeng Li

hp-l33

Efficient AI

published a dataset about 3 hours ago

updated a dataset about 3 hours ago

updated a model about 10 hours ago

hp-l33 's Spaces 6

FP8 sparse attention with a compressed tail

Qwen-Image 2.1 4-step LoRAs + W8A8 + TAE on a 24 GB M5 Mac

RTX 5090 优化报告：约 2.05× 加速、质量 gate 与可下载实测证据

Explore side‑by‑side video comparisons of diffusion experiments

Generate high‑resolution videos from text prompts

LTX-2.5 Stage 2 Sol-Attn latency and quality on 4xB200.