FastDeploy

test_activation.py

[Feature] use phi permute/unpermute & rm swiglu (#6361 )

2026-03-12 02:01:57 -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

[Feature]Supports SWA based on appendattn (#6547 )

2026-03-01 19:02:08 +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

[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 )

2026-03-24 11:12:39 +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_attention_versions_determinism.py

[Feature] Add Deterministic Inference Support (#6476 )

2026-02-26 19:31:51 -08:00

test_flash_attn_func.py

fix fa4 test (#6408 )

2026-02-10 10:57:21 +08:00

test_fp8_ue8m0.py

[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 )

2026-03-24 11:12:39 +08:00

test_fused_moe_cutlass_backend.py

[Feature] Support MOE Cutlass backend for latent MOE (#7428 )

2026-04-16 22:11:49 +08:00

test_fused_moe_triton_backend.py

[Feature] 添加 MoE 层 latent mode 支持 (#7382 )

2026-04-15 13:57:07 +08:00

test_fusedmoe.py

[Feature] 添加 MoE 层 latent mode 支持 (#7382 )

2026-04-15 13:57:07 +08:00

test_guided_decoding.py

[Others]update paddleformer 1.0.0 (#6496 )

2026-03-11 15:06:29 +08:00

test_kv_cache_int8_dynamic_quant_backend.py

Split enable_mm (#7183 )

2026-04-08 11:25:41 +08:00

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_nvfp4_fusedmoe.py

[Feature] 为 FusedMoE 添加 hidden_size 显式参数支持 (#7361 )

2026-04-13 20:24:58 +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 )

2026-02-24 15:20:29 +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

[Cleanup] Replace torch proxy alias with public compat API (#7348 )

2026-04-13 11:43:26 +08:00

test_speculative_sampler.py

[Optimization][Speculative Decoding]Fuse padding sampling params (#6765 )

2026-03-12 05:05:15 -07:00

test_trtllm_allreduce_rms_fusion.py

[Optimization] enable trtllm_all_reduce fusion kernel in glm model (#6660 )

2026-04-16 14:10:19 +08:00

test_w4a8_moe.py

[Feature] 为 FusedMoE 添加 hidden_size 显式参数支持 (#7361 )

2026-04-13 20:24:58 +08:00

test_w4afp8_moe.py

[Feature] 为 FusedMoE 添加 hidden_size 显式参数支持 (#7361 )

2026-04-13 20:24:58 +08:00

trtllm_allreduce_rms_fusion.py

[Optimization] enable trtllm_all_reduce fusion kernel in glm model (#6660 )

2026-04-16 14:10:19 +08:00