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.
17 lines
469 B
Python
17 lines
469 B
Python
"""
|
|
Reasoning parser module for vLLM 0.6.3 (BI-V100 / Qwen3.6-27B adaptation).
|
|
|
|
Usage: --reasoning-parser qwen3
|
|
"""
|
|
|
|
from vllm.reasoning.abs_reasoning_parsers import ReasoningParser, ReasoningParserManager
|
|
|
|
__all__ = ["ReasoningParser", "ReasoningParserManager"]
|
|
|
|
# Lazy-register Qwen3 parser; imported on first get_reasoning_parser("qwen3").
|
|
ReasoningParserManager.register_lazy(
|
|
"qwen3",
|
|
"vllm.reasoning.qwen3_reasoning_parser",
|
|
"Qwen3ReasoningParser",
|
|
)
|