wangxiyuan c8b0f5f799 [4/N][Refactor] torchair model runner refactor (#2208) ...

There is lot of torchair code in model runner leading the code hard for
maintenance. We'll create new torchair_model_runner to split torchair
related logic. Following the workflow #2203, this is the first PR.

What's this PR do:

create common function `_convert_torch_foramt`  for initialize_kv_cache


- vLLM version: v0.10.0
- vLLM main:
14a5d903ab

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>

2025-08-11 21:39:24 +08:00

..

attention

[Feat] chunkprefill mla support torchair graph (#1772)

2025-08-11 19:58:59 +08:00

compilation

[CI] Upgrade vllm to 0.9.1 (#1165)

2025-06-11 16:33:11 +08:00

core

Disaggregate prefill for kv cache register style (#950)

2025-07-26 17:15:47 +08:00

device_allocator

[Misc]Clean up useless import from vllm (#2049)

2025-07-28 16:01:59 +08:00

distributed

[Misc] Nit fix for disaggregated_prefill and ascend_forward_context (#2097)

2025-08-05 08:39:02 +08:00

lora

[Bugfix] LoRA logits einsum dimension mismatch in add_lora_logits (#1583)

2025-07-30 09:50:36 +08:00

models

【main】SP For Qwen3 MoE (#2209)

2025-08-07 09:15:49 +08:00

multistream

[MISC] Cherry pick #1291 from v0.9.1-dev (#1825)

2025-08-01 09:08:45 +08:00

ops

[CI][Quickfix] Fix AscendFusedMoE init error (#2268)

2025-08-08 10:20:23 +08:00

patch

enable mm allreduce test (#2192)

2025-08-07 17:19:23 +08:00

quantization

[V1] MTP supports torchair (#2145)

2025-08-06 19:37:43 +08:00

sample

[Perf][MTP] Optimize reject sampler in greedy situation. (#2137)

2025-08-11 17:37:49 +08:00

torchair

[4/N][Refactor] torchair model runner refactor (#2208)

2025-08-11 21:39:24 +08:00

worker

[4/N][Refactor] torchair model runner refactor (#2208)

2025-08-11 21:39:24 +08:00

__init__.py

Bump torch version to 2.7.1 (#1562)

2025-08-05 08:43:24 +08:00

ascend_config.py

[Doc] Support kimi-k2-w8a8 (#2162)

2025-08-06 19:28:47 +08:00

ascend_forward_context.py

[Misc] Nit fix for disaggregated_prefill and ascend_forward_context (#2097)

2025-08-05 08:39:02 +08:00

envs.py

[main][Feature]Moe alltoallv communication optimization for unquantized RL training sence (#2088)

2025-08-02 09:49:10 +08:00

meta_registration.py

[core] Support capture custom ops into aclgraph (#2113)

2025-08-11 15:59:42 +08:00

platform.py

【main】SP For Qwen3 MoE (#2209)

2025-08-07 09:15:49 +08:00

utils.py

[core] Support capture custom ops into aclgraph (#2113)

2025-08-11 15:59:42 +08:00