chen 47fbf1401a [Cherry-Pick][RL]moe bf16 ep support paddle batch_gemm(#7337) (#7338) ...

* moe bf16 ep support paddle batch_gemm

2026-04-11 21:51:19 +08:00

..

test_activation.py

[Cherry-Pick][Feature] use phi permute/unpermute & rm swiglu (#6361) (#6808)

2026-03-13 08:00:22 -07:00

test_append_attention_with_output.py

[Others]get_block_shape_and_split_kv_block clean code (#5123)

2025-11-20 16:40:04 +08:00

test_append_attention.py

[Cherry-Pick][Feature]Supports SWA based on appendattn #6547 (#6594)

2026-03-02 20:15:23 +08:00

test_attention_layer.py

[Feature] consider multimodal model when dummy run (#6045)

2026-02-09 17:49:55 +08:00

test_batched_count_greater_than.py

[Optimization] Optimization for gather_logprob by 10GB (#5817)

2025-12-30 15:33:34 +08:00

test_deepgemm_fused_moe.py

[Cherry-Pick][RL] Support moe_topk_select using Paddle native operators and Add fused stack-transpose-quant for BlockWiseFP8 MoE weight quantization and swiglu-fp8-quant op for DeepGemmFusedMoE for training alignment (#6850) (#6935)

2026-03-24 11:12:51 +08:00

test_ep_moe_expert_dispatch_fp8.py

[Others] remove template NUM_EXPERTS_PER_RANK in permute_x_fp8_kernel (#5620)

2026-01-04 11:21:15 +08:00

test_ffn.py

[Models] Add forward_meta to moe models' forward function (#5138)

2025-12-04 13:26:58 +08:00

test_flash_attn_func.py

[Cherry-Pick][CI]fix fa4 test (#6408)(#6424) (#6421)

2026-02-10 20:38:55 +08:00

test_fp8_ue8m0.py

[Cherry-Pick][RL] Support moe_topk_select using Paddle native operators and Add fused stack-transpose-quant for BlockWiseFP8 MoE weight quantization and swiglu-fp8-quant op for DeepGemmFusedMoE for training alignment (#6850) (#6935)

2026-03-24 11:12:51 +08:00

test_fused_moe_cutlass_backend.py

[Cherry-Pick][RL]moe bf16 ep support paddle batch_gemm(#7337) (#7338)

2026-04-11 21:51:19 +08:00

test_fused_moe_triton_backend.py

[Cherry-Pick][RL] Support moe_topk_select using Paddle native operators and Add fused stack-transpose-quant for BlockWiseFP8 MoE weight quantization and swiglu-fp8-quant op for DeepGemmFusedMoE for training alignment (#6850) (#6935)

2026-03-24 11:12:51 +08:00

test_fusedmoe.py

[UNITEST] add EP TP test_fused_moe CI (#5989)

2026-01-15 21:37:32 +08:00

test_guided_decoding.py

[Cherry-Pick][Others]update paddleformer 1.0.0 #6496 (#6752)

2026-03-27 01:24:22 -07:00

test_min_sampling.py

Add stable ci (#3460)

2025-08-20 08:57:17 +08:00

test_moba_attention_backend.py

[KVCache] support unified cache backend (#4903)

2025-11-12 14:54:52 +08:00

test_native_paddle_backend.py

[Docs] Add License in Unittest (#4957)

2025-11-12 10:44:09 +08:00

test_plas_attention.py

[Executor] Default use CUDAGraph (#3594)

2025-10-21 14:25:45 +08:00

test_qkvg_parallel_linear.py

[Feature] support qkv&gate linear fusion (#6455) (#6552)

2026-03-03 11:32:13 +08:00

test_quantized_linear.py

[Optimization] Support FA2/FA3/FA4 with attn_mask_q (#6354)

2026-02-05 14:39:00 +08:00

test_repetition_early_stopper.py

Add stable ci (#3460)

2025-08-20 08:57:17 +08:00

test_sampler.py

[Cherry-Pick][Others]update paddleformer 1.0.0 #6496 (#6752)

2026-03-27 01:24:22 -07:00

test_speculative_sampler.py

[Feature] consider multimodal model when dummy run (#6045)

2026-02-09 17:49:55 +08:00

test_w4a8_moe.py

[Optimization] Support FA2/FA3/FA4 with attn_mask_q (#6354)

2026-02-05 14:39:00 +08:00

test_w4afp8_moe.py

[Optimization] Support FA2/FA3/FA4 with attn_mask_q (#6354)

2026-02-05 14:39:00 +08:00