Vadim Gimpelson 717ac33d9c

[PERF] Qwen3-next. Add fp8 cutlass MoE tuned configs. chmod -x *MI308X.json (#29553) ...

Signed-off-by: Vadim Gimpelson <vadim.gimpelson@gmail.com>

2025-12-18 13:16:04 +08:00

..

layers

[PERF] Qwen3-next. Add fp8 cutlass MoE tuned configs. chmod -x *MI308X.json (#29553)

2025-12-18 13:16:04 +08:00

model_loader

[Model][Quantization] Restore MoE + GGUF models support (incl. Qwen3 MoE) by allowing Sideload Parameters (#30116)

2025-12-09 05:30:05 +00:00

models

[refactor] Add prefix support to embed_tokens in DeepSeek MTP (#30788)

2025-12-18 04:45:56 +00:00

warmup

[UX] Reduce DeepGEMM warmup log output to single progress bar (#30903)

2025-12-17 20:21:51 -08:00

__init__.py

…

custom_op.py

[CustomOp] Support object-level enable for CustomOp (#30547)

2025-12-15 11:02:09 +08:00

parameter.py

…

utils.py

[Quantization] FP8 Weight Reloading for Quantized RL Rollout (#28480)

2025-12-09 13:54:32 -08:00