project6-dev 7969874cde fix: cast router_logits to float32 before topk kernel ...

router_logits from ReplicatedLinear is float16. PyTorch path does
.float() but kernel path passed float16 directly. BI-V100 corex CUB
half-precision softmax produces garbage. Cast to float32 first.

2026-08-12 10:06:19 +00:00

..

prebuilt/corex-3.2.3-ivcore10

prebuilt: add corex_moe_topk_softmax.so (verified on BI-V100)

2026-08-12 07:11:25 +00:00

qwen3_5

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

qwen3_5_moe

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

reasoning

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

vendor_overrides/vllm

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

wheels

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

api_server.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

bi100_env.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

bi100_profile.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

block_major_kv_cache.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

build_corex_attn_head_rms_norm.sh

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

build_corex_fused_paged_prefill_split4.sh

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

build_corex_gdn_beta_decay.sh

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

build_corex_gdn_causal_conv.sh

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

build_corex_gdn_gated_norm.sh

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

build_corex_gdn_packed_decode.sh

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

build_corex_gdn_qk_map.sh

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

build_corex_moe_direct_routed.sh

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

build_corex_moe_exact_reduce.sh

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

build_corex_moe_topk_softmax.sh

feat(moe): corex_moe_topk_softmax — CUB-based fused topk+softmax kernel

2026-08-12 06:15:15 +00:00

build_corex_moe_weight_gather.sh

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

build_corex_paged_kv_gather.sh

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

chat_utils.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

cli_args.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

corex_attn_head_rms_norm.cu

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

corex_block_major_kv_transfer.cu

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

corex_fused_paged_prefill_split4.cu

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

corex_gdn_beta_decay.cu

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

corex_gdn_causal_conv.cu

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

corex_gdn_gated_norm.cu

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

corex_gdn_packed_decode.cu

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

corex_gdn_qk_map.cu

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

corex_moe_direct_routed.cu

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

corex_moe_exact_reduce.cu

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

corex_moe_topk_softmax.cu

feat(moe): corex_moe_topk_softmax — CUB-based fused topk+softmax kernel

2026-08-12 06:15:15 +00:00

corex_moe_weight_gather.cu

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

corex_paged_kv_gather.cu

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

corex_query_tiled_paged_prefill.cu

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

device_utils.cuh

feat(moe): corex_moe_topk_softmax — CUB-based fused topk+softmax kernel

2026-08-12 06:15:15 +00:00

gdn_prefix.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

install_prebuilt_corex.sh

fix: prebuilt expects 13 .so (add corex_moe_topk_softmax slot)

2026-08-12 06:22:36 +00:00

mamba_cache.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

moe_topk_sigmoid_kernels.cuh

fix(build): remove cuda/functional include — not available on corex

2026-08-12 06:42:54 +00:00

moe_topk_softmax_kernels.cuh

fix(build): properly convert all glog CHECK/LOG to TORCH_CHECK syntax

2026-08-12 06:57:33 +00:00

paged_attn.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_block_major_cache_engine.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_block_major_worker_capacity.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_block_manager_cache_trace.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_corex_swap_blocks.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_executor_startup_debug.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_model_runner.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_ops.sh

fix: prebuilt expects 13 .so (add corex_moe_topk_softmax slot)

2026-08-12 06:22:36 +00:00

patch_transformers_qwen3_5.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_utils.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_vllm_qwen3_5.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_vllm_tool_parser.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_worker_cache_transfer_order.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_worker_profile_override.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_worker_startup_profile_guard.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_xformers_profile.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_xformers_sdpa_batch_kernel.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_xformers_sdpa_batch.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_xformers_sdpa_seq_kernel.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

patch_xformers_sdpa_seq.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

protocol.py

fix: accept max_completion_tokens + reasoning_effort + n=2

2026-08-12 06:16:44 +00:00

qwen3_5.py

fix: cast router_logits to float32 before topk kernel

2026-08-12 10:06:19 +00:00

qwen3coder_tool_parser.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

scheduler.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

sequence.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00

serving_chat.py

fix: accept max_completion_tokens + reasoning_effort + n=2

2026-08-12 06:16:44 +00:00

serving_tokenization.py

feat: 替换为 project_7 验证通过的 wudixzy stack

2026-08-12 03:31:05 +00:00