Levi 9862a23985 【0.11.0-dev】optimization of kimi-k2 in cann8.3 (#4555) ...

### What this PR does / why we need it?
In cann8.3， npu_moe_gating_top_k operator can support expert nums with
384, so kimi can use the operator to get better preformance.
---------

Signed-off-by: Levi-JQ <yujinqi2@huawei.com>
Co-authored-by: Levi-JQ <yujinqi2@huawei.com>

2025-12-09 08:49:15 +08:00

..

models

[BugFix] Fix torchair+mtp bug after deleting deepseek_mtp. (#3590)

2025-10-21 22:23:52 +08:00

ops

【0.11.0-dev】optimization of kimi-k2 in cann8.3 (#4555)

2025-12-09 08:49:15 +08:00

quantization

【0.11.0-dev】optimization of kimi-k2 in cann8.3 (#4555)

2025-12-09 08:49:15 +08:00

__init__.py

[1/4][Refactor] Refactor torchair worker (#1885)

2025-07-21 11:50:46 +08:00

torchair_attention.py

[v0.11.0][Perf] Eliminating the zerolike operator through patch (#3632)

2025-10-23 14:49:28 +08:00

torchair_mla.py

[0.11.0][BugFix] Improve the performance of prefixcache features (#4021)

2025-11-10 11:51:34 +08:00

torchair_model_runner.py

fix bug when max_seqs=14 in mtp=2 scenario and raise error when cudagraph_capture_sizes can't be an integer multiple of uniform_decode_query_len (#3909)

2025-10-31 09:25:06 +08:00

torchair_sfa.py

For nz unset in bf16&fp16 (#4495)

2025-11-28 17:32:25 +08:00

torchair_worker.py

[CI] Upgrade vllm to newest commit (#3182)

2025-09-26 06:18:15 +08:00

utils.py

For nz unset in bf16&fp16 (#4495)

2025-11-28 17:32:25 +08:00