ef6abf3dc7
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.
43 lines
1.4 KiB
Python
43 lines
1.4 KiB
Python
from typing import List, Optional, Union
|
|
|
|
from vllm.logger import init_logger
|
|
from vllm.lora.request import LoRARequest
|
|
from vllm.pooling_params import PoolingParams
|
|
from vllm.prompt_adapter.request import PromptAdapterRequest
|
|
from vllm.sampling_params import BeamSearchParams, SamplingParams
|
|
|
|
logger = init_logger(__name__)
|
|
|
|
|
|
class RequestLogger:
|
|
|
|
def __init__(self, *, max_log_len: Optional[int]) -> None:
|
|
super().__init__()
|
|
|
|
self.max_log_len = max_log_len
|
|
|
|
def log_inputs(
|
|
self,
|
|
request_id: str,
|
|
prompt: Optional[str],
|
|
prompt_token_ids: Optional[List[int]],
|
|
params: Optional[Union[SamplingParams, PoolingParams,
|
|
BeamSearchParams]],
|
|
lora_request: Optional[LoRARequest],
|
|
prompt_adapter_request: Optional[PromptAdapterRequest],
|
|
) -> None:
|
|
max_log_len = self.max_log_len
|
|
if max_log_len is not None:
|
|
if prompt is not None:
|
|
prompt = prompt[:max_log_len]
|
|
|
|
if prompt_token_ids is not None:
|
|
prompt_token_ids = prompt_token_ids[:max_log_len]
|
|
|
|
logger.info(
|
|
"Received request %s: prompt: %r, "
|
|
"params: %s, prompt_token_ids: %s, "
|
|
"lora_request: %s, prompt_adapter_request: %s.", request_id,
|
|
prompt, params, prompt_token_ids, lora_request,
|
|
prompt_adapter_request)
|