lzy 223b2f5d86 Support setting communication groups in custom_allreduce and the all-to-all\transpose fused operator during the decoding phase. (#5917)

2026-01-12 14:09:39 +08:00

..

graph_optimization

[Others] add assert and only count the actual load in cuda_graph (#5445)

2025-12-10 11:22:54 +08:00

guided_decoding

[Feature] Guided Decoding add LLguidance backend (#5124)

2025-12-03 20:23:57 +08:00

layers

Support setting communication groups in custom_allreduce and the all-to-all\transpose fused operator during the decoding phase. (#5917)

2026-01-12 14:09:39 +08:00

logits_processor

[Feature] support logits processors (#4515)

2025-10-29 00:08:53 +08:00

model_loader

[Loader]Fix bug in MTP weight loading (#5744)

2025-12-25 11:32:17 +08:00

models

[Feature] Support redundant expert for eplb (#5918)

2026-01-09 17:13:24 +08:00

ops

[INTEL HPU] support only one release package of PaddleCustomDevice (#5910)

2026-01-08 11:57:13 +08:00

__init__.py

polish code with new pre-commit rule (#2923)

2025-07-19 23:19:27 +08:00

entropy_utils.py

[Bugfix] Fix entropy calculation bugs (#5941)

2026-01-08 20:57:35 +08:00

forward_meta.py

[Intel HPU] enable chunked prefill (#5903)

2026-01-06 21:01:50 +08:00

load_weight_utils.py

[Feature] Support redundant expert for eplb (#5918)

2026-01-09 17:13:24 +08:00

pre_and_post_process.py

[Metax] optimize flash attention backend (#5876)

2026-01-06 09:52:09 +08:00

utils.py

[INTEL_HPU] supported ERNIE-4.5-21B-A3B-Thinking (#5891)

2026-01-07 21:31:53 +08:00

xpu_pre_and_post_process.py

[XPU] fix dp4 (#5946)

2026-01-09 20:36:53 +08:00