[CP]Glm45 air 2.2 (#4073)

* [Feature] Support zai-org/GLM-4.5-Air BF16 model (#3928)

* support glm45_air

* [Feature] GLM-45-AIR Support Mix Quantization(Dense wfp8afp8 and wint8 triton_moe_backend) (#4051)

* check

* fix v1 load for mix and wint8

* check --quantizations 'None'

* check

* support RL rollout

* check v1 loader

* check glm rollout_model, change wfp8afp8 per_token_cast_to_fp8 to native impl

* check rollout moe gate begin layer_id

* check rollout e_score_correction_bias

* delete infer_to_train_mapping={}

* code check

fastdeploy/rl/rollout_config.py

@@ -14,6 +14,8 @@
 # limitations under the License.
 """
 
+from typing import Any, Dict, Optional
+
 from fastdeploy.worker.worker_process import initialize_fd_config
 
 
@@ -52,7 +54,7 @@ class RolloutModelConfig:
         expert_parallel_size: int = 1,
         enable_expert_parallel: bool = False,
         ori_vocab_size: int = None,
-        quantization: str = "None",
+        quantization: Optional[Dict[str, Any]] = None,
         guided_decoding_backend: str = "off",
         disable_any_whitespace: bool = True,
         enable_logprob: bool = False,