Aryan Kumar cc9bfd069d [perf]: Load H3 DiT weights onto CUDA, not CPU ...

GB10 unified memory made the CPU-then-copy path take minutes for
the 35B FastH3 transformer. When host offload is off, read
safetensors onto the accelerator.

2026-08-30 23:27:24 -07:00

..

benchmarks

[misc] yapf pass under CI's interpreter (3.12) + pin hook language_version (#1702)

2026-08-08 22:24:17 -07:00

__init__.py

[Feature] Remove V1 folder (#642)

2025-07-24 22:43:12 -07:00

component_loader.py

[feat] Add MiniMax H3 LoRA inference and preview launchers (#1771)

2026-08-27 14:43:39 -07:00

fsdp_load.py

[perf]: Load H3 DiT weights onto CUDA, not CPU

2026-08-30 23:27:24 -07:00

lora_patch.py

[feat] Add MiniMax H3 LoRA inference and preview launchers (#1771)

2026-08-27 14:43:39 -07:00

text_encoder_quantization.py

[perf] Optimize MiniMax-H3 text encoder memory (#1732)

2026-08-21 14:57:06 -07:00

utils.py

[misc] yapf pass under CI's interpreter (3.12) + pin hook language_version (#1702)

2026-08-08 22:24:17 -07:00

weight_utils.py

[misc] yapf pass under CI's interpreter (3.12) + pin hook language_version (#1702)

2026-08-08 22:24:17 -07:00