ROCm: bump 6.3.0 (#3259)

python/pyproject.toml

@@ -19,31 +19,29 @@ dependencies = ["requests", "tqdm", "numpy", "IPython", "setproctitle"]
 runtime_common = [
     "aiohttp", "decord", "fastapi",
     "hf_transfer", "huggingface_hub", "interegular", "modelscope",
-    "orjson", "outlines>=0.0.44,<0.1.0",
-    "packaging", "pillow", "prometheus-client>=0.20.0",
-    "psutil", "pydantic", "python-multipart",
-    "pyzmq>=25.1.2", "torchao>=0.7.0", "uvicorn", "uvloop",
-    "xgrammar>=0.1.10"
+    "orjson", "packaging", "pillow", "prometheus-client>=0.20.0",
+    "psutil", "pydantic", "python-multipart", "pyzmq>=25.1.2",
+    "torchao>=0.7.0", "uvicorn", "uvloop", "xgrammar>=0.1.10"
 ]
 srt = [
     "sglang[runtime_common]", "cuda-python",
     "sgl-kernel>=0.0.3.post1", "torch", "vllm==0.6.4.post1",
-    "flashinfer==0.1.6"
+    "flashinfer==0.1.6", "outlines>=0.0.44,<0.1.0"
 ]
 
 # HIP (Heterogeneous-computing Interface for Portability) for AMD
 # => base docker rocm/vllm-dev:20241022, not from public vllm whl
-srt_hip = ["sglang[runtime_common]", "torch", "vllm==0.6.3.post2.dev1"]
+srt_hip = ["sglang[runtime_common]", "torch", "vllm==0.6.7.dev2", "outlines==0.1.11"]
 # xpu is not enabled in public vllm and torch whl,
 # need to follow https://docs.vllm.ai/en/latest/getting_started/xpu-installation.htmlinstall vllm
-srt_xpu = ["sglang[runtime_common]"]
+srt_xpu = ["sglang[runtime_common]", "outlines>=0.0.44,<0.1.0"]
 #For Intel Gaudi(device : hpu) follow the installation guide
 #https://docs.vllm.ai/en/latest/getting_started/gaudi-installation.html
-srt_hpu = ["sglang[runtime_common]"]
+srt_hpu = ["sglang[runtime_common]", "outlines>=0.0.44,<0.1.0"]
 # CPU: currently, there are no pre-built vllm wheels for CPU.
 # To install vllm for CPU, please follow the instruction here:
 # https://docs.vllm.ai/en/latest/getting_started/installation/cpu/index.html
-srt_cpu = ["sglang[runtime_common]", "torch"]
+srt_cpu = ["sglang[runtime_common]", "torch", "outlines>=0.0.44,<0.1.0"]
 
 openai = ["openai>=1.0", "tiktoken"]
 anthropic = ["anthropic>=0.20.0"]

python/sglang/srt/constrained/outlines_backend.py

@@ -20,7 +20,6 @@ from typing import Dict, List, Optional, Tuple, Union
 import interegular
 import torch
 from outlines.fsm.guide import RegexGuide
-from outlines.fsm.json_schema import build_regex_from_schema
 from outlines.models.transformers import TransformerTokenizer
 from pydantic import BaseModel
 
@@ -29,6 +28,15 @@ from sglang.srt.constrained.base_grammar_backend import (
     BaseGrammarObject,
 )
 from sglang.srt.constrained.outlines_jump_forward import OutlinesJumpForwardMap
+from sglang.srt.utils import is_hip
+
+is_hip_ = is_hip()
+
+if is_hip_:
+    from outlines_core.fsm.json_schema import build_regex_from_schema
+else:
+    from outlines.fsm.json_schema import build_regex_from_schema
+
 
 logger = logging.getLogger(__name__)
 

python/sglang/srt/custom_op.py

@@ -20,7 +20,7 @@ class CustomOp(nn.Module):
         raise NotImplementedError
 
     def forward_hip(self, *args, **kwargs):
-        raise NotImplementedError
+        return self.forward_native(*args, **kwargs)
 
     def forward_xpu(self, *args, **kwargs):
         return self.forward_native(*args, **kwargs)