[DEPLOY] Complete submission: baseline + all optimizations

Adds ALL files needed for Dockerfile build:
  - qwen3_6_scripts/ (baseline patches + our optimizations)
  - vllm/ (full vllm package)
  - paged_attention_v2_pytorch.py (V2 with single-bmm optimization)
  - Dockerfile + computility-run.yaml

Our optimizations vs baseline:
  1. paged_attn.py: pre-gathered context KV (eliminates 194 gather calls),
     Triton try/fallback, V2 heuristic, threshold 32K→64K
  2. paged_attention_v2_pytorch.py: fills NotImplementedError,
     single-bmm Phase 1 (195 launches → 3)
  3. patch_enable_triton.py: HAS_TRITON=True with safety fallback
  4. patch_triton_tuning.py: BLOCK=64, NUM_WARPS=4 for BI-V100
  5. computility-run.yaml: gpu-memory-utilization 0.9→0.95,
     max-num-batched-tokens 8192→16384

This repo can now be submitted to dev.modelhub.org.cn as-is.

vllm/transformers_utils/configs/eagle.py

@@ -0,0 +1,49 @@
+import os
+from typing import Optional, Union
+
+from transformers import AutoConfig, PretrainedConfig
+
+
+class EAGLEConfig(PretrainedConfig):
+    model_type = "eagle"
+
+    def __init__(self,
+                 model: Union[PretrainedConfig, dict, None] = None,
+                 truncated_vocab_size: Optional[int] = None,
+                 **kwargs):
+
+        model_config = None if model is None else (AutoConfig.for_model(
+            **model) if isinstance(model, dict) else model)
+
+        for k, v in kwargs.items():
+            if k != "architectures" and k != "model_type" and hasattr(
+                    model_config, k):
+                setattr(model_config, k, v)
+
+        self.model = model_config
+
+        if self.model is None:
+            self.truncated_vocab_size = None
+        else:
+            self.truncated_vocab_size = self.model.vocab_size if \
+                truncated_vocab_size is None else truncated_vocab_size
+
+        if "architectures" not in kwargs:
+            kwargs["architectures"] = ["EAGLEModel"]
+
+        super().__init__(**kwargs)
+
+        if self.model is not None:
+            for k, v in self.model.to_dict().items():
+                if not hasattr(self, k):
+                    setattr(self, k, v)
+
+    @classmethod
+    def from_pretrained(
+        cls,
+        pretrained_model_name_or_path: Union[str, os.PathLike],
+        **kwargs,
+    ) -> "EAGLEConfig":
+        config_dict, kwargs = cls.get_config_dict(
+            pretrained_model_name_or_path, **kwargs)
+        return cls.from_dict(config_dict, **kwargs)