File size: 10,696 Bytes
31dc8dc | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 | from __future__ import annotations
import argparse
from dataclasses import dataclass
from typing import Sequence
def parse_device_ids(value: str | None) -> list[int]:
if value is None or value.strip() == "":
return []
return [int(item.strip()) for item in value.split(",") if item.strip()]
@dataclass
class ServerArgs:
model: str
host: str = "0.0.0.0"
port: int = 8000
log_level: str = "info"
zmq_command_addr: str = ""
zmq_event_addr: str = ""
model_name: str = "dream"
decoding_strategy: str = "d2f"
sampling_mode: str = "naive"
tensor_parallel_size: int = 1
data_parallel_size: int = 1
master_addr: str = "localhost"
master_port: int = 2333
distributed_timeout_seconds: int = 600
device_ids: list[int] | None = None
block_size: int = 32
buffer_size: int = 4
page_size: int = 32
max_num_batched_tokens: int = 4096
max_num_reqs: int = 128
max_model_len: int = 2048
enable_prefill_cudagraph: bool = True
enable_full_static_runner: bool = True
prefill_cudagraph_max_len: int = 0
enable_torch_compile: bool = True
enable_cudagraph_torch_compile: bool = False
torch_compile_mode: str = "reduce-overhead"
auto_max_nfe_warmup_steps: int = 8
auto_max_nfe_tpf_floor: float = 1.0
gpu_memory_utilization: float = 0.9
enforce_eager: bool = False
attn_impl: str = "triton"
enable_prefix_caching: bool = True
kv_cache_layout: str = "unified"
moe_dispatcher_backend: str = "standard"
moe_gemm_impl: str = "triton"
deepep_mode: str = "auto"
deepep_num_max_dispatch_tokens_per_rank: int = 256
add_block_threshold: float | None = None
semi_complete_threshold: float | None = None
accept_threshold: float | None = None
remask_threshold: float | None = None
token_stability_threshold: float | None = None
use_lora: bool = False
lora_path: str = ""
pre_merge_lora: bool = False
def engine_kwargs(self) -> dict:
return {
"model_name": self.model_name,
"decoding_strategy": self.decoding_strategy,
"sampling_mode": self.sampling_mode,
"tensor_parallel_size": self.tensor_parallel_size,
"data_parallel_size": self.data_parallel_size,
"master_addr": self.master_addr,
"master_port": self.master_port,
"distributed_timeout_seconds": self.distributed_timeout_seconds,
"device_ids": self.device_ids or [],
"block_size": self.block_size,
"buffer_size": self.buffer_size,
"page_size": self.page_size,
"max_num_batched_tokens": self.max_num_batched_tokens,
"max_num_reqs": self.max_num_reqs,
"max_model_len": self.max_model_len,
"enable_prefill_cudagraph": self.enable_prefill_cudagraph,
"enable_full_static_runner": self.enable_full_static_runner,
"prefill_cudagraph_max_len": self.prefill_cudagraph_max_len,
"enable_torch_compile": self.enable_torch_compile,
"enable_cudagraph_torch_compile": self.enable_cudagraph_torch_compile,
"torch_compile_mode": self.torch_compile_mode,
"auto_max_nfe_warmup_steps": self.auto_max_nfe_warmup_steps,
"auto_max_nfe_tpf_floor": self.auto_max_nfe_tpf_floor,
"gpu_memory_utilization": self.gpu_memory_utilization,
"enforce_eager": self.enforce_eager,
"attn_impl": self.attn_impl,
"enable_prefix_caching": self.enable_prefix_caching,
"kv_cache_layout": self.kv_cache_layout,
"moe_dispatcher_backend": self.moe_dispatcher_backend,
"moe_gemm_impl": self.moe_gemm_impl,
"deepep_mode": self.deepep_mode,
"deepep_num_max_dispatch_tokens_per_rank": self.deepep_num_max_dispatch_tokens_per_rank,
"decoding_thresholds": {
"add_block_threshold": 0.1 if self.add_block_threshold is None else self.add_block_threshold,
"semi_complete_threshold": 0.9
if self.semi_complete_threshold is None
else self.semi_complete_threshold,
"accept_threshold": 0.9 if self.accept_threshold is None else self.accept_threshold,
"remask_threshold": 0.4 if self.remask_threshold is None else self.remask_threshold,
"token_stability_threshold": 0.0
if self.token_stability_threshold is None
else self.token_stability_threshold,
},
"use_lora": self.use_lora,
"lora_path": self.lora_path,
"pre_merge_lora": self.pre_merge_lora,
}
def build_arg_parser() -> argparse.ArgumentParser:
parser = argparse.ArgumentParser(description="Launch a Diffulex HTTP server")
parser.add_argument("--model", required=True, help="Path to the local model directory")
parser.add_argument("--host", default="0.0.0.0")
parser.add_argument("--port", type=int, default=8000)
parser.add_argument("--log-level", default="info")
parser.add_argument("--zmq-command-addr", default="", help="Frontend-to-backend ZMQ PUSH/PULL address")
parser.add_argument("--zmq-event-addr", default="", help="Backend-to-frontend ZMQ PUSH/PULL address")
parser.add_argument("--model-name", default="dream")
parser.add_argument("--decoding-strategy", default="d2f")
parser.add_argument("--sampling-mode", default="naive", choices=["naive", "edit"])
parser.add_argument("--tensor-parallel-size", type=int, default=1)
parser.add_argument("--data-parallel-size", type=int, default=1)
parser.add_argument("--master-addr", default="localhost")
parser.add_argument("--master-port", type=int, default=2333)
parser.add_argument("--distributed-timeout-seconds", type=int, default=600)
parser.add_argument("--device-ids", default="", help="Comma-separated logical CUDA device ids")
parser.add_argument("--block-size", type=int, default=32)
parser.add_argument("--buffer-size", type=int, default=4)
parser.add_argument("--page-size", type=int, default=32)
parser.add_argument("--max-num-batched-tokens", type=int, default=4096)
parser.add_argument("--max-num-reqs", type=int, default=128)
parser.add_argument("--max-model-len", type=int, default=2048)
parser.add_argument("--disable-prefill-cudagraph", action="store_true")
parser.add_argument("--disable-full-static-runner", action="store_true")
parser.add_argument("--prefill-cudagraph-max-len", type=int, default=0)
parser.add_argument("--disable-torch-compile", action="store_true")
parser.add_argument("--enable-cudagraph-torch-compile", action="store_true")
parser.add_argument("--torch-compile-mode", default="reduce-overhead")
parser.add_argument("--auto-max-nfe-warmup-steps", type=int, default=8)
parser.add_argument("--auto-max-nfe-tpf-floor", type=float, default=1.0)
parser.add_argument("--gpu-memory-utilization", type=float, default=0.9)
parser.add_argument("--enforce-eager", action="store_true")
parser.add_argument("--attn-impl", default="triton", choices=["triton", "naive"])
parser.add_argument("--disable-prefix-caching", action="store_true")
parser.add_argument("--kv-cache-layout", default="unified", choices=["unified", "distinct"])
parser.add_argument("--moe-dispatcher-backend", default="standard", choices=["standard", "naive", "deepep"])
parser.add_argument("--moe-gemm-impl", default="triton", choices=["triton", "vllm", "vllm_modular", "naive"])
parser.add_argument("--deepep-mode", default="auto", choices=["normal", "low_latency", "auto"])
parser.add_argument("--deepep-num-max-dispatch-tokens-per-rank", type=int, default=256)
parser.add_argument("--add-block-threshold", type=float, default=None)
parser.add_argument("--semi-complete-threshold", type=float, default=None)
parser.add_argument("--accept-threshold", type=float, default=None)
parser.add_argument("--remask-threshold", type=float, default=None)
parser.add_argument("--token-stability-threshold", type=float, default=None)
parser.add_argument("--use-lora", action="store_true")
parser.add_argument("--lora-path", default="")
parser.add_argument("--pre-merge-lora", action="store_true")
return parser
def parse_args(argv: Sequence[str] | None = None) -> ServerArgs:
ns = build_arg_parser().parse_args(argv)
return ServerArgs(
model=ns.model,
host=ns.host,
port=ns.port,
log_level=ns.log_level,
zmq_command_addr=ns.zmq_command_addr,
zmq_event_addr=ns.zmq_event_addr,
model_name=ns.model_name,
decoding_strategy=ns.decoding_strategy,
sampling_mode=ns.sampling_mode,
tensor_parallel_size=ns.tensor_parallel_size,
data_parallel_size=ns.data_parallel_size,
master_addr=ns.master_addr,
master_port=ns.master_port,
distributed_timeout_seconds=ns.distributed_timeout_seconds,
device_ids=parse_device_ids(ns.device_ids),
block_size=ns.block_size,
buffer_size=ns.buffer_size,
page_size=ns.page_size,
max_num_batched_tokens=ns.max_num_batched_tokens,
max_num_reqs=ns.max_num_reqs,
max_model_len=ns.max_model_len,
enable_prefill_cudagraph=not ns.disable_prefill_cudagraph,
enable_full_static_runner=not ns.disable_full_static_runner,
prefill_cudagraph_max_len=ns.prefill_cudagraph_max_len,
enable_torch_compile=not ns.disable_torch_compile,
enable_cudagraph_torch_compile=ns.enable_cudagraph_torch_compile,
torch_compile_mode=ns.torch_compile_mode,
auto_max_nfe_warmup_steps=ns.auto_max_nfe_warmup_steps,
auto_max_nfe_tpf_floor=ns.auto_max_nfe_tpf_floor,
gpu_memory_utilization=ns.gpu_memory_utilization,
enforce_eager=ns.enforce_eager,
attn_impl=ns.attn_impl,
enable_prefix_caching=not ns.disable_prefix_caching,
kv_cache_layout=ns.kv_cache_layout,
moe_dispatcher_backend=ns.moe_dispatcher_backend,
moe_gemm_impl=ns.moe_gemm_impl,
deepep_mode=ns.deepep_mode,
deepep_num_max_dispatch_tokens_per_rank=ns.deepep_num_max_dispatch_tokens_per_rank,
add_block_threshold=ns.add_block_threshold,
semi_complete_threshold=ns.semi_complete_threshold,
accept_threshold=ns.accept_threshold,
remask_threshold=ns.remask_threshold,
token_stability_threshold=ns.token_stability_threshold,
use_lora=ns.use_lora,
lora_path=ns.lora_path,
pre_merge_lora=ns.pre_merge_lora,
)
|