希望添加对 Qwen3-Omni 的量化工具

Describe the feature
当前项目提供的量化工具无法对 Qwen3-Omni 进行量化

Paste any useful information
我尝试使用了位于 https://github.com/modelscope/ms-swift/blob/main/examples/export/quantize/omni/gptq.sh 的量化脚本，但看起来这个对于 Qwen2.5-Omni 的量化脚本不可用，因为 Qwen3-Omni 使用了全新的 Thinker-Talker 架构，使用该脚本会报错：
AssertionError: mllm_arch.language_model: ['thinker.model', 'thinker.lm_head']

我也尝试使用了 llm-compressor 项目进行量化，但会有如下报错：

ValueError: Unrecognized configuration class <class 'transformers.models.qwen3_omni_moe.configuration_qwen3_omni_moe.Qwen3OmniMoeConfig'> for this kind of AutoModel: AutoModelForCausalLM.
Model type should be one of ApertusConfig, ArceeConfig, AriaTextConfig, BambaConfig, BartConfig, BertConfig, BertGenerationConfig, BigBirdConfig, BigBirdPegasusConfig, BioGptConfig, BitNetConfig, BlenderbotConfig, BlenderbotSmallConfig, BloomConfig, BltConfig, CamembertConfig, LlamaConfig, CodeGenConfig, CohereConfig, Cohere2Config, CpmAntConfig, CTRLConfig, Data2VecTextConfig, DbrxConfig, DeepseekV2Config, DeepseekV3Config, DiffLlamaConfig, DogeConfig, Dots1Config, ElectraConfig, Emu3Config, ErnieConfig, Ernie4_5Config, Ernie4_5_MoeConfig, Exaone4Config, FalconConfig, FalconH1Config, FalconMambaConfig, FlexOlmoConfig, FuyuConfig, GemmaConfig, Gemma2Config, Gemma3Config, Gemma3TextConfig, Gemma3nConfig, Gemma3nTextConfig, GitConfig, GlmConfig, Glm4Config, Glm4MoeConfig, GotOcr2Config, GPT2Config, GPT2Config, GPTBigCodeConfig, GPTNeoConfig, GPTNeoXConfig, GPTNeoXJapaneseConfig, GptOssConfig, GPTJConfig, GraniteConfig, GraniteMoeConfig, GraniteMoeHybridConfig, GraniteMoeSharedConfig, HeliumConfig, HunYuanDenseV1Config, HunYuanMoEV1Config, JambaConfig, JetMoeConfig, Lfm2Config, LlamaConfig, Llama4Config, Llama4TextConfig, LongcatFlashConfig, MambaConfig, Mamba2Config, MarianConfig, MBartConfig, MegaConfig, MegatronBertConfig, MiniMaxConfig, MinistralConfig, MistralConfig, MixtralConfig, MllamaConfig, ModernBertDecoderConfig, MoshiConfig, MptConfig, MusicgenConfig, MusicgenMelodyConfig, MvpConfig, NemotronConfig, OlmoConfig, Olmo2Config, Olmo3Config, OlmoeConfig, OpenLlamaConfig, OpenAIGPTConfig, OPTConfig, PegasusConfig, PersimmonConfig, PhiConfig, Phi3Config, Phi4MultimodalConfig, PhimoeConfig, PLBartConfig, ProphetNetConfig, QDQBertConfig, Qwen2Config, Qwen2MoeConfig, Qwen3Config, Qwen3MoeConfig, Qwen3NextConfig, RecurrentGemmaConfig, ReformerConfig, RemBertConfig, RobertaConfig, RobertaPreLayerNormConfig, RoCBertConfig, RoFormerConfig, RwkvConfig, SeedOssConfig, SmolLM3Config, Speech2Text2Config, StableLmConfig, Starcoder2Config, TransfoXLConfig, TrOCRConfig, VaultGemmaConfig, WhisperConfig, XGLMConfig, XLMConfig, XLMProphetNetConfig, XLMRobertaConfig, XLMRobertaXLConfig, XLNetConfig, xLSTMConfig, XmodConfig, ZambaConfig, Zamba2Config.

若使用 llm-compressor 量化，最接近 Qwen3-Omni 的是 Qwen3MoeConfig ，但仍然没有提供 Qwen3-Omni 的选项。所以看起来 llm-compressor 项目也没有对 Qwen3-Omni 进行适配

希望增加对于 Qwen3-Omni 已微调完毕权重的量化工具支持，感谢！

Start by comparing the existing examples/export/quantize/omni/gptq.sh flow for Qwen2.5-Omni with the Qwen3-Omni Thinker-Talker architecture. Reproduce the reported assertion and inspect how llm-compressor handles Qwen3MoeConfig and AutoModelForCausalLM. Done means a documented quantization path for fine-tuned Qwen3-Omni weights without these configuration errors.