701ae503
tangwang
docs
|
1
2
3
4
5
6
7
8
9
10
11
|
"""
Qwen3-Reranker-0.6B backend using vLLM.
Reference: https://huggingface.co/Qwen/Qwen3-Reranker-0.6B
Requires: vllm>=0.8.5, transformers; GPU recommended.
"""
from __future__ import annotations
import logging
import math
|
9f5994b4
tangwang
reranker
|
12
|
import os
|
efd435cf
tangwang
tei性能调优:
|
13
|
import threading
|
701ae503
tangwang
docs
|
14
|
import time
|
9f5994b4
tangwang
reranker
|
15
16
|
from typing import Any, Dict, List, Tuple
|
701ae503
tangwang
docs
|
17
18
|
logger = logging.getLogger("reranker.backends.qwen3_vllm")
|
3d588bef
tangwang
embeddings
|
19
20
21
22
|
import torch
from transformers import AutoTokenizer
from vllm import LLM, SamplingParams
from vllm.inputs.data import TokensPrompt
|
701ae503
tangwang
docs
|
23
24
|
|
985752f5
tangwang
1. 前端调试功能
|
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
|
def deduplicate_with_positions(texts: List[str]) -> Tuple[List[str], List[int]]:
"""
Deduplicate texts globally while preserving first-seen order.
Returns:
unique_texts: deduplicated texts in first-seen order
position_to_unique: mapping from each original position to unique index
"""
unique_texts: List[str] = []
position_to_unique: List[int] = []
seen: Dict[str, int] = {}
for text in texts:
idx = seen.get(text)
if idx is None:
idx = len(unique_texts)
seen[text] = idx
unique_texts.append(text)
position_to_unique.append(idx)
return unique_texts, position_to_unique
|
701ae503
tangwang
docs
|
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
|
def _format_instruction(instruction: str, query: str, doc: str) -> List[Dict[str, str]]:
"""Build chat messages for one (query, doc) pair."""
return [
{
"role": "system",
"content": "Judge whether the Document meets the requirements based on the Query and the Instruct provided. Note that the answer can only be \"yes\" or \"no\".",
},
{
"role": "user",
"content": f"<Instruct>: {instruction}\n\n<Query>: {query}\n\n<Document>: {doc}",
},
]
class Qwen3VLLMRerankerBackend:
"""
Qwen3-Reranker-0.6B with vLLM inference.
Config from services.rerank.backends.qwen3_vllm.
"""
def __init__(self, config: Dict[str, Any]) -> None:
self._config = config or {}
model_name = str(self._config.get("model_name") or "Qwen/Qwen3-Reranker-0.6B")
|
07cf5a93
tangwang
START_EMBEDDING=...
|
71
|
max_model_len = int(self._config.get("max_model_len", 2048))
|
701ae503
tangwang
docs
|
72
|
tensor_parallel_size = int(self._config.get("tensor_parallel_size", 1))
|
07cf5a93
tangwang
START_EMBEDDING=...
|
73
74
75
76
|
gpu_memory_utilization = float(self._config.get("gpu_memory_utilization", 0.4))
enable_prefix_caching = bool(self._config.get("enable_prefix_caching", False))
enforce_eager = bool(self._config.get("enforce_eager", True))
dtype = str(self._config.get("dtype", "float16")).strip().lower()
|
701ae503
tangwang
docs
|
77
78
|
self._instruction = str(
self._config.get("instruction")
|
fb973d19
tangwang
configs
|
79
|
or "Given a query, score the product for relevance"
|
701ae503
tangwang
docs
|
80
|
)
|
9f5994b4
tangwang
reranker
|
81
82
83
84
|
infer_batch_size = os.getenv("RERANK_VLLM_INFER_BATCH_SIZE") or self._config.get("infer_batch_size", 64)
sort_by_doc_length = os.getenv("RERANK_VLLM_SORT_BY_DOC_LENGTH")
if sort_by_doc_length is None:
sort_by_doc_length = self._config.get("sort_by_doc_length", True)
|
9f5994b4
tangwang
reranker
|
85
86
87
|
self._infer_batch_size = int(infer_batch_size)
self._sort_by_doc_length = str(sort_by_doc_length).strip().lower() in {"1", "true", "yes", "y", "on"}
|
07cf5a93
tangwang
START_EMBEDDING=...
|
88
89
90
91
|
if not torch.cuda.is_available():
raise RuntimeError("qwen3_vllm backend requires CUDA GPU, but torch.cuda.is_available() is False")
if dtype not in {"float16", "half", "auto"}:
raise ValueError(f"Unsupported dtype for qwen3_vllm: {dtype!r}. Use float16/half/auto.")
|
9f5994b4
tangwang
reranker
|
92
|
if self._infer_batch_size <= 0:
|
985752f5
tangwang
1. 前端调试功能
|
93
94
95
|
raise ValueError(
f"infer_batch_size must be > 0, got {self._infer_batch_size}"
)
|
701ae503
tangwang
docs
|
96
97
|
logger.info(
|
07cf5a93
tangwang
START_EMBEDDING=...
|
98
|
"[Qwen3_VLLM] Loading model %s (max_model_len=%s, tp=%s, gpu_mem=%.2f, dtype=%s, prefix_caching=%s)",
|
701ae503
tangwang
docs
|
99
100
101
|
model_name,
max_model_len,
tensor_parallel_size,
|
07cf5a93
tangwang
START_EMBEDDING=...
|
102
103
|
gpu_memory_utilization,
dtype,
|
701ae503
tangwang
docs
|
104
105
106
107
108
109
110
111
112
|
enable_prefix_caching,
)
self._llm = LLM(
model=model_name,
tensor_parallel_size=tensor_parallel_size,
max_model_len=max_model_len,
gpu_memory_utilization=gpu_memory_utilization,
enable_prefix_caching=enable_prefix_caching,
|
07cf5a93
tangwang
START_EMBEDDING=...
|
113
114
|
enforce_eager=enforce_eager,
dtype=dtype,
|
701ae503
tangwang
docs
|
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
|
)
self._tokenizer = AutoTokenizer.from_pretrained(model_name)
self._tokenizer.padding_side = "left"
self._tokenizer.pad_token = self._tokenizer.eos_token
# Suffix for generation prompt (assistant answer)
self._suffix = "<|im_end|>\n<|im_start|>assistant\n<think>\n\n</think>\n\n"
self._suffix_tokens = self._tokenizer.encode(
self._suffix, add_special_tokens=False
)
self._max_prompt_len = max_model_len - len(self._suffix_tokens)
self._true_token = self._tokenizer("yes", add_special_tokens=False).input_ids[0]
self._false_token = self._tokenizer("no", add_special_tokens=False).input_ids[0]
self._sampling_params = SamplingParams(
temperature=0,
max_tokens=1,
logprobs=20,
allowed_token_ids=[self._true_token, self._false_token],
)
|
efd435cf
tangwang
tei性能调优:
|
135
136
137
|
# vLLM generate path is unstable under concurrent calls in this process model.
# Serialize infer calls to avoid engine-core protocol corruption.
self._infer_lock = threading.Lock()
|
701ae503
tangwang
docs
|
138
139
140
141
142
143
144
145
|
self._model_name = model_name
logger.info("[Qwen3_VLLM] Model ready | model=%s", model_name)
def _process_inputs(
self,
pairs: List[Tuple[str, str]],
) -> List[TokensPrompt]:
|
bc089b43
tangwang
refactor(reranker...
|
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
|
"""Build tokenized prompts for vLLM from (query, doc) pairs. Batch apply_chat_template."""
messages_batch = [
_format_instruction(self._instruction, q, d) for q, d in pairs
]
tokenized = self._tokenizer.apply_chat_template(
messages_batch,
tokenize=True,
add_generation_prompt=False,
enable_thinking=False,
)
# Single conv returns flat list; batch returns list of lists
if tokenized and not isinstance(tokenized[0], list):
tokenized = [tokenized]
prompts = [
TokensPrompt(
prompt_token_ids=ids[: self._max_prompt_len] + self._suffix_tokens
|
701ae503
tangwang
docs
|
162
|
)
|
bc089b43
tangwang
refactor(reranker...
|
163
164
|
for ids in tokenized
]
|
701ae503
tangwang
docs
|
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
|
return prompts
def _compute_scores(
self,
prompts: List[TokensPrompt],
) -> List[float]:
"""Run vLLM generate and compute yes/no probability per prompt."""
if not prompts:
return []
outputs = self._llm.generate(prompts, self._sampling_params, use_tqdm=False)
scores = []
for i in range(len(outputs)):
out = outputs[i]
if not out.outputs:
scores.append(0.0)
continue
|
bc089b43
tangwang
refactor(reranker...
|
181
182
|
final_logits = out.outputs[0].logprobs
if not final_logits:
|
701ae503
tangwang
docs
|
183
184
|
scores.append(0.0)
continue
|
bc089b43
tangwang
refactor(reranker...
|
185
186
187
188
189
190
191
192
193
194
195
196
197
|
last = final_logits[-1]
# Match official: missing token -> logprob = -10
if self._true_token not in last:
true_logit = -10
else:
true_logit = last[self._true_token].logprob
if self._false_token not in last:
false_logit = -10
else:
false_logit = last[self._false_token].logprob
true_score = math.exp(true_logit)
false_score = math.exp(false_logit)
score = true_score / (true_score + false_score)
|
701ae503
tangwang
docs
|
198
199
200
|
scores.append(float(score))
return scores
|
9f5994b4
tangwang
reranker
|
201
202
203
204
205
206
207
|
def _estimate_doc_lengths(self, docs: List[str]) -> List[int]:
"""
Estimate token lengths for sorting documents into similar-length batches.
Falls back to character length when tokenizer length output is unavailable.
"""
if not docs:
return []
|
985752f5
tangwang
1. 前端调试功能
|
208
|
# Use simple character length to approximate document length.
|
9f5994b4
tangwang
reranker
|
209
210
|
return [len(text) for text in docs]
|
701ae503
tangwang
docs
|
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
|
def score_with_meta(
self,
query: str,
docs: List[str],
normalize: bool = True,
) -> Tuple[List[float], Dict[str, Any]]:
start_ts = time.time()
total_docs = len(docs) if docs else 0
output_scores: List[float] = [0.0] * total_docs
query = "" if query is None else str(query).strip()
indexed: List[Tuple[int, str]] = []
for i, doc in enumerate(docs or []):
if doc is None:
continue
text = str(doc).strip()
if not text:
continue
indexed.append((i, text))
if not query or not indexed:
elapsed_ms = (time.time() - start_ts) * 1000.0
return output_scores, {
"input_docs": total_docs,
"usable_docs": len(indexed),
"unique_docs": 0,
"dedup_ratio": 0.0,
"elapsed_ms": round(elapsed_ms, 3),
"model": self._model_name,
"backend": "qwen3_vllm",
"normalize": normalize,
|
9f5994b4
tangwang
reranker
|
242
243
244
|
"infer_batch_size": self._infer_batch_size,
"inference_batches": 0,
"sort_by_doc_length": self._sort_by_doc_length,
|
701ae503
tangwang
docs
|
245
246
|
}
|
9f5994b4
tangwang
reranker
|
247
248
249
250
251
252
253
|
# Deduplicate globally by text, keep mapping to original indices.
indexed_texts = [text for _, text in indexed]
unique_texts, position_to_unique = deduplicate_with_positions(indexed_texts)
lengths = self._estimate_doc_lengths(unique_texts)
order = list(range(len(unique_texts)))
if self._sort_by_doc_length and len(unique_texts) > 1:
|
985752f5
tangwang
1. 前端调试功能
|
254
|
order = sorted(order, key=lambda i: lengths[i])
|
9f5994b4
tangwang
reranker
|
255
256
257
|
unique_scores: List[float] = [0.0] * len(unique_texts)
inference_batches = 0
|
985752f5
tangwang
1. 前端调试功能
|
258
259
|
for start in range(0, len(order), self._infer_batch_size):
batch_indices = order[start : start + self._infer_batch_size]
|
9f5994b4
tangwang
reranker
|
260
261
|
inference_batches += 1
pairs = [(query, unique_texts[i]) for i in batch_indices]
|
efd435cf
tangwang
tei性能调优:
|
262
|
prompts = self._process_inputs(pairs)
|
9f5994b4
tangwang
reranker
|
263
264
265
266
267
268
269
270
|
with self._infer_lock:
batch_scores = self._compute_scores(prompts)
if len(batch_scores) != len(batch_indices):
raise RuntimeError(
f"Reranker score size mismatch: expected {len(batch_indices)}, got {len(batch_scores)}"
)
for idx, score in zip(batch_indices, batch_scores):
unique_scores[idx] = float(score)
|
701ae503
tangwang
docs
|
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
|
for (orig_idx, _), unique_idx in zip(indexed, position_to_unique):
# Score is already P(yes) in [0,1] from yes/(yes+no)
output_scores[orig_idx] = float(unique_scores[unique_idx])
elapsed_ms = (time.time() - start_ts) * 1000.0
dedup_ratio = 0.0
if indexed:
dedup_ratio = 1.0 - (len(unique_texts) / float(len(indexed)))
meta = {
"input_docs": total_docs,
"usable_docs": len(indexed),
"unique_docs": len(unique_texts),
"dedup_ratio": round(dedup_ratio, 4),
"elapsed_ms": round(elapsed_ms, 3),
"model": self._model_name,
"backend": "qwen3_vllm",
"normalize": normalize,
|
9f5994b4
tangwang
reranker
|
290
291
|
"infer_batch_size": self._infer_batch_size,
"inference_batches": inference_batches,
|
af827ce9
tangwang
rerank
|
292
|
"sort_by_doc_length": self._sort_by_doc_length
|
701ae503
tangwang
docs
|
293
294
|
}
return output_scores, meta
|