Aryan Kumar 5bc3c0902d [feat]: optional Theseus warm-up for the FastH3 QAD method ...

theseus_steps > 0 trains teacher-forced segment reconstruction first, widening
from single blocks to the whole stack (JustQuant, arXiv:2609.33601), then hands off
to the configured objective.

2026-10-30 01:15:45 -07:00

..

__init__.py

[feat] pre-commit support 120 col num (#1167)

2026-03-11 20:19:30 -07:00

kd.py

[feat] pre-commit support 120 col num (#1167)

2026-03-11 20:19:30 -07:00

pdd_qad_metrics.py

[feat]: add the shared FastH3 NVFP4 FP4-attention QAD recipe

2026-10-19 12:00:33 -07:00

pdd_qad_recon.py

[feat]: optional Theseus warm-up for the FastH3 QAD method

2026-10-30 01:15:45 -07:00

pdd_qad_step0.py

[fix]: address QAD review findings

2026-10-19 12:16:45 -07:00

pdd_qad.py

[feat]: optional Theseus warm-up for the FastH3 QAD method

2026-10-30 01:15:45 -07:00