[MetaxGPU] Support FastDeploy on metax gpu (#3241)

* [MetaxGPU] Support FastDeploy on metax gpu

* Update metax_worker.py

1. change worker log;
2. remove custom allreduce, adapt it later;
3. remove cuda graph;

* Update __init__.py

1. remove metax's key work comment

* Update __init__.py

1. remove metax's key word comment;
2. add fused_moe_kernel_paddle import

---------

Co-authored-by: yongqiangma <xing.wo@163.com>

fastdeploy/model_executor/layers/sample/ops/apply_penalty_multi_scores.py

@@ -119,6 +119,23 @@ def apply_penalty_multi_scores(
             min_dec_lens,
             eos_token_ids,
         )
+    elif current_platform.is_maca():
+        from fastdeploy.model_executor.ops.gpu import get_token_penalty_multi_scores
+
+        logits = get_token_penalty_multi_scores(
+            pre_token_ids,
+            prompt_ids,
+            prompt_lens,
+            logits,
+            repetition_penalties,
+            frequency_penalties,
+            presence_penalties,
+            temperature,
+            bad_words_token_ids,
+            step_idx,
+            min_dec_lens,
+            eos_token_ids,
+        )
     else:
         raise NotImplementedError
 

fastdeploy/model_executor/layers/sample/sampler.py

@@ -177,6 +177,7 @@ class Sampler(nn.Layer):
             or current_platform.is_iluvatar()
             or current_platform.is_gcu()
             or current_platform.is_dcu()
+            or current_platform.is_maca()
         ):
             self.forward = self.forward_cuda
         else: