TommyAI1986 commited on
Commit
4e2c7b4
·
verified ·
1 Parent(s): cb05fc2

Add Qwen3 Reranker 0.6B (Core AI)

Browse files
.gitattributes CHANGED
@@ -35,3 +35,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
  qwen3-embedding-0.6b/qwen3-embedding-0.6b_float16_s512_static.aimodel/main.mlirb filter=lfs diff=lfs merge=lfs -text
37
  qwen3-embedding-0.6b/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 
 
 
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
  qwen3-embedding-0.6b/qwen3-embedding-0.6b_float16_s512_static.aimodel/main.mlirb filter=lfs diff=lfs merge=lfs -text
37
  qwen3-embedding-0.6b/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
38
+ qwen3-reranker-0.6b/qwen3-reranker-0.6b_float16_s512_static.aimodel/main.mlirb filter=lfs diff=lfs merge=lfs -text
39
+ qwen3-reranker-0.6b/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
qwen3-reranker-0.6b/SHA256SUMS ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 5bba8c734f6dd3ae48317b4139317e45a7fce48fc55e15670b23a0dd15492ab6 SOURCE-README.md
2
+ 5ccfcb4fc6a54a89ee8f3f701ef9e5e18d80b4edd79ae3f4631994b4fecadac1 artifact-manifest.json
3
+ 88206a3757a27a746bc47dda42cc00e1a1331e4da6ac5c40eb52cf2685104d34 compatibility-marker.json
4
+ e046edbb8576942d0cd2ec3ddb07a4aa8428284c76624619ffdb03f82d8724bd export-metadata.json
5
+ 44bc199b2defb36d55b13ecfd10c798db0c3b14b8f01cdf4c99a36a5887f2215 installation-receipt.json
6
+ f6ee2245aadd6d1353600343997fe05ec943ea1c69242ed8c728acc827be0f25 parity-report.json
7
+ 134bec116ce705528d66f47ce9e6d8d3228e28ecc04380c7bd74358bdf9e97d6 qwen3-reranker-0.6b_float16_s512_static.aimodel/main.hash
8
+ 5978b9117f2a2b9e40bed4135aaad5cce976e8e31f042e228ba0dadff720e714 qwen3-reranker-0.6b_float16_s512_static.aimodel/main.mlirb
9
+ 8782f6a734c4f215acead4ad6ac2cd215105368a49fdd57f8586fc48befbcf0c qwen3-reranker-0.6b_float16_s512_static.aimodel/metadata.json
10
+ 134e3ab21fd252452fcde142ff1dd4de0c52ddb64f93f4e1820c2babe15745a0 reference.json
11
+ e2e83e9605959c4af840284cacb4aa328cade64dcd26d3dbb8a90a5372d442ce reranker-contract.json
12
+ 849145e0dbf4b9665b8623b908f846a1223f713aff98974b8e8fd6af8e194fa8 source-revision-receipt.json
13
+ 975b0852c1198287773194cfbacb7d9595124fe43a481f48171c4067ba3d4917 swift-tokenizer-golden.json
14
+ 6f682162495ec5b39fd9005c01b6aa2a74669379fe967039f1e2cbbe8752369d tokenizer/chat_template.jinja
15
+ be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506 tokenizer/tokenizer.json
16
+ 779410a34a98fb2c6c925a1be6f24d632f1723473161b037f6676eb6c04886c6 tokenizer/tokenizer_config.json
17
+ 19070dbaadbc07baacf868f75b4d0c51cf6d4bfb4df9e34f960ee306690f12a0 validation-report.json
qwen3-reranker-0.6b/SOURCE-README.md ADDED
@@ -0,0 +1,291 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model:
4
+ - Qwen/Qwen3-0.6B-Base
5
+ library_name: transformers
6
+ tags:
7
+ - sentence-transformers
8
+ pipeline_tag: text-ranking
9
+ ---
10
+ # Qwen3-Reranker-0.6B
11
+
12
+ <p align="center">
13
+ <img src="https://qianwen-res.oss-accelerate-overseas.aliyuncs.com/logo_qwen3.png" width="400"/>
14
+ <p>
15
+
16
+ ## Highlights
17
+
18
+ The Qwen3 Embedding model series is the latest proprietary model of the Qwen family, specifically designed for text embedding and ranking tasks. Building upon the dense foundational models of the Qwen3 series, it provides a comprehensive range of text embeddings and reranking models in various sizes (0.6B, 4B, and 8B). This series inherits the exceptional multilingual capabilities, long-text understanding, and reasoning skills of its foundational model. The Qwen3 Embedding series represents significant advancements in multiple text embedding and ranking tasks, including text retrieval, code retrieval, text classification, text clustering, and bitext mining.
19
+
20
+ **Exceptional Versatility**: The embedding model has achieved state-of-the-art performance across a wide range of downstream application evaluations. The 8B size embedding model ranks No.1 in the MTEB multilingual leaderboard (as of June 5, 2025, score 70.58), while the reranking model excels in various text retrieval scenarios.
21
+
22
+ **Comprehensive Flexibility**: The Qwen3 Embedding series offers a full spectrum of sizes (from 0.6B to 8B) for both embedding and reranking models, catering to diverse use cases that prioritize efficiency and effectiveness. Developers can seamlessly combine these two modules. Additionally, the embedding model allows for flexible vector definitions across all dimensions, and both embedding and reranking models support user-defined instructions to enhance performance for specific tasks, languages, or scenarios.
23
+
24
+ **Multilingual Capability**: The Qwen3 Embedding series offer support for over 100 languages, thanks to the multilingual capabilites of Qwen3 models. This includes various programming languages, and provides robust multilingual, cross-lingual, and code retrieval capabilities.
25
+
26
+ ## Model Overview
27
+
28
+ **Qwen3-Reranker-0.6B** has the following features:
29
+
30
+ - Model Type: Text Reranking
31
+ - Supported Languages: 100+ Languages
32
+ - Number of Paramaters: 0.6B
33
+ - Context Length: 32k
34
+
35
+ For more details, including benchmark evaluation, hardware requirements, and inference performance, please refer to our [blog](https://qwenlm.github.io/blog/qwen3-embedding/), [GitHub](https://github.com/QwenLM/Qwen3-Embedding).
36
+
37
+ ## Qwen3 Embedding Series Model list
38
+
39
+ | Model Type | Models | Size | Layers | Sequence Length | Embedding Dimension | MRL Support | Instruction Aware |
40
+ |------------------|----------------------|------|--------|-----------------|---------------------|-------------|----------------|
41
+ | Text Embedding | [Qwen3-Embedding-0.6B](https://huggingface.co/Qwen/Qwen3-Embedding-0.6B) | 0.6B | 28 | 32K | 1024 | Yes | Yes |
42
+ | Text Embedding | [Qwen3-Embedding-4B](https://huggingface.co/Qwen/Qwen3-Embedding-4B) | 4B | 36 | 32K | 2560 | Yes | Yes |
43
+ | Text Embedding | [Qwen3-Embedding-8B](https://huggingface.co/Qwen/Qwen3-Embedding-8B) | 8B | 36 | 32K | 4096 | Yes | Yes |
44
+ | Text Reranking | [Qwen3-Reranker-0.6B](https://huggingface.co/Qwen/Qwen3-Reranker-0.6B) | 0.6B | 28 | 32K | - | - | Yes |
45
+ | Text Reranking | [Qwen3-Reranker-4B](https://huggingface.co/Qwen/Qwen3-Reranker-4B) | 4B | 36 | 32K | - | - | Yes |
46
+ | Text Reranking | [Qwen3-Reranker-8B](https://huggingface.co/Qwen/Qwen3-Reranker-8B) | 8B | 36 | 32K | - | - | Yes |
47
+
48
+ > **Note**:
49
+ > - `MRL Support` indicates whether the embedding model supports custom dimensions for the final embedding.
50
+ > - `Instruction Aware` notes whether the embedding or reranking model supports customizing the input instruction according to different tasks.
51
+ > - Our evaluation indicates that, for most downstream tasks, using instructions (instruct) typically yields an improvement of 1% to 5% compared to not using them. Therefore, we recommend that developers create tailored instructions specific to their tasks and scenarios. In multilingual contexts, we also advise users to write their instructions in English, as most instructions utilized during the model training process were originally written in English.
52
+
53
+
54
+ ## Usage
55
+
56
+ ### Using Sentence Transformers
57
+
58
+ Install Sentence Transformers:
59
+ ```bash
60
+ pip install sentence_transformers
61
+ ```
62
+
63
+ ```python
64
+ from sentence_transformers import CrossEncoder
65
+
66
+ model = CrossEncoder("Qwen/Qwen3-Reranker-0.6B")
67
+
68
+ query = "What is the capital of China?"
69
+ documents = [
70
+ "The capital of China is Beijing.",
71
+ "Gravity is a force that attracts two bodies towards each other. It gives weight to physical objects and is responsible for the movement of planets around the sun.",
72
+ ]
73
+
74
+ pairs = [(query, doc) for doc in documents]
75
+ scores = model.predict(pairs)
76
+ print(scores)
77
+ # [ 7.625 -11.375]
78
+
79
+ rankings = model.rank(query, documents)
80
+ print(rankings)
81
+ # [{'corpus_id': 0, 'score': 7.625}, {'corpus_id': 1, 'score': -11.375}]
82
+ ```
83
+
84
+ By default, scores are raw logit differences. To get 0-1 probability scores, pass a Sigmoid activation function:
85
+ ```python
86
+ scores = model.predict([(query, doc) for doc in documents], activation_fn=torch.nn.Sigmoid())
87
+ ```
88
+
89
+ The model uses a default prompt `"query"` which injects the instruction `"Given a web search query, retrieve relevant passages that answer the query"` into the chat template. You can provide a custom instruction via the `prompts` parameter:
90
+ ```python
91
+ model = CrossEncoder(
92
+ "Qwen/Qwen3-Reranker-0.6B",
93
+ prompts={"classification": "Classify whether the document matches the query topic"},
94
+ default_prompt_name="classification",
95
+ )
96
+ ```
97
+
98
+ ### Using Transformers
99
+
100
+ With Transformers versions earlier than 4.51.0, you may encounter the following error:
101
+ ```
102
+ KeyError: 'qwen3'
103
+ ```
104
+
105
+ ```python
106
+ # Requires transformers>=4.51.0
107
+ import torch
108
+ from transformers import AutoModel, AutoTokenizer, AutoModelForCausalLM
109
+
110
+ def format_instruction(instruction, query, doc):
111
+ if instruction is None:
112
+ instruction = 'Given a web search query, retrieve relevant passages that answer the query'
113
+ output = "<Instruct>: {instruction}\n<Query>: {query}\n<Document>: {doc}".format(instruction=instruction,query=query, doc=doc)
114
+ return output
115
+
116
+ def process_inputs(pairs):
117
+ inputs = tokenizer(
118
+ pairs, padding=False, truncation='longest_first',
119
+ return_attention_mask=False, max_length=max_length - len(prefix_tokens) - len(suffix_tokens)
120
+ )
121
+ for i, ele in enumerate(inputs['input_ids']):
122
+ inputs['input_ids'][i] = prefix_tokens + ele + suffix_tokens
123
+ inputs = tokenizer.pad(inputs, padding=True, return_tensors="pt", max_length=max_length)
124
+ for key in inputs:
125
+ inputs[key] = inputs[key].to(model.device)
126
+ return inputs
127
+
128
+ @torch.no_grad()
129
+ def compute_logits(inputs, **kwargs):
130
+ batch_scores = model(**inputs).logits[:, -1, :]
131
+ true_vector = batch_scores[:, token_true_id]
132
+ false_vector = batch_scores[:, token_false_id]
133
+ batch_scores = torch.stack([false_vector, true_vector], dim=1)
134
+ batch_scores = torch.nn.functional.log_softmax(batch_scores, dim=1)
135
+ scores = batch_scores[:, 1].exp().tolist()
136
+ return scores
137
+
138
+ tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-Reranker-0.6B", padding_side='left')
139
+ model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-Reranker-0.6B").eval()
140
+ # We recommend enabling flash_attention_2 for better acceleration and memory saving.
141
+ # model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-Reranker-0.6B", torch_dtype=torch.float16, attn_implementation="flash_attention_2").cuda().eval()
142
+ token_false_id = tokenizer.convert_tokens_to_ids("no")
143
+ token_true_id = tokenizer.convert_tokens_to_ids("yes")
144
+ max_length = 8192
145
+
146
+ prefix = "<|im_start|>system\nJudge whether the Document meets the requirements based on the Query and the Instruct provided. Note that the answer can only be \"yes\" or \"no\".<|im_end|>\n<|im_start|>user\n"
147
+ suffix = "<|im_end|>\n<|im_start|>assistant\n<think>\n\n</think>\n\n"
148
+ prefix_tokens = tokenizer.encode(prefix, add_special_tokens=False)
149
+ suffix_tokens = tokenizer.encode(suffix, add_special_tokens=False)
150
+
151
+ task = 'Given a web search query, retrieve relevant passages that answer the query'
152
+
153
+ queries = ["What is the capital of China?",
154
+ "Explain gravity",
155
+ ]
156
+
157
+ documents = [
158
+ "The capital of China is Beijing.",
159
+ "Gravity is a force that attracts two bodies towards each other. It gives weight to physical objects and is responsible for the movement of planets around the sun.",
160
+ ]
161
+
162
+ pairs = [format_instruction(task, query, doc) for query, doc in zip(queries, documents)]
163
+
164
+ # Tokenize the input texts
165
+ inputs = process_inputs(pairs)
166
+ scores = compute_logits(inputs)
167
+
168
+ print("scores: ", scores)
169
+ ```
170
+
171
+
172
+ ### vLLM Usage
173
+
174
+ ```python
175
+ # Requires vllm>=0.8.5
176
+ import logging
177
+ from typing import Dict, Optional, List
178
+
179
+ import json
180
+ import logging
181
+
182
+ import torch
183
+
184
+ from transformers import AutoTokenizer, is_torch_npu_available
185
+ from vllm import LLM, SamplingParams
186
+ from vllm.distributed.parallel_state import destroy_model_parallel
187
+ import gc
188
+ import math
189
+ from vllm.inputs.data import TokensPrompt
190
+
191
+
192
+
193
+ def format_instruction(instruction, query, doc):
194
+ text = [
195
+ {"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\"."},
196
+ {"role": "user", "content": f"<Instruct>: {instruction}\n\n<Query>: {query}\n\n<Document>: {doc}"}
197
+ ]
198
+ return text
199
+
200
+ def process_inputs(pairs, instruction, max_length, suffix_tokens):
201
+ messages = [format_instruction(instruction, query, doc) for query, doc in pairs]
202
+ messages = tokenizer.apply_chat_template(
203
+ messages, tokenize=True, add_generation_prompt=False, enable_thinking=False
204
+ )
205
+ messages = [ele[:max_length] + suffix_tokens for ele in messages]
206
+ messages = [TokensPrompt(prompt_token_ids=ele) for ele in messages]
207
+ return messages
208
+
209
+ def compute_logits(model, messages, sampling_params, true_token, false_token):
210
+ outputs = model.generate(messages, sampling_params, use_tqdm=False)
211
+ scores = []
212
+ for i in range(len(outputs)):
213
+ final_logits = outputs[i].outputs[0].logprobs[-1]
214
+ token_count = len(outputs[i].outputs[0].token_ids)
215
+ if true_token not in final_logits:
216
+ true_logit = -10
217
+ else:
218
+ true_logit = final_logits[true_token].logprob
219
+ if false_token not in final_logits:
220
+ false_logit = -10
221
+ else:
222
+ false_logit = final_logits[false_token].logprob
223
+ true_score = math.exp(true_logit)
224
+ false_score = math.exp(false_logit)
225
+ score = true_score / (true_score + false_score)
226
+ scores.append(score)
227
+ return scores
228
+
229
+ number_of_gpu = torch.cuda.device_count()
230
+ tokenizer = AutoTokenizer.from_pretrained('Qwen/Qwen3-Reranker-0.6B')
231
+ model = LLM(model='Qwen/Qwen3-Reranker-0.6B', tensor_parallel_size=number_of_gpu, max_model_len=10000, enable_prefix_caching=True, gpu_memory_utilization=0.8)
232
+ tokenizer.padding_side = "left"
233
+ tokenizer.pad_token = tokenizer.eos_token
234
+ suffix = "<|im_end|>\n<|im_start|>assistant\n<think>\n\n</think>\n\n"
235
+ max_length=8192
236
+ suffix_tokens = tokenizer.encode(suffix, add_special_tokens=False)
237
+ true_token = tokenizer("yes", add_special_tokens=False).input_ids[0]
238
+ false_token = tokenizer("no", add_special_tokens=False).input_ids[0]
239
+ sampling_params = SamplingParams(temperature=0,
240
+ max_tokens=1,
241
+ logprobs=20,
242
+ allowed_token_ids=[true_token, false_token],
243
+ )
244
+
245
+
246
+ task = 'Given a web search query, retrieve relevant passages that answer the query'
247
+ queries = ["What is the capital of China?",
248
+ "Explain gravity",
249
+ ]
250
+ documents = [
251
+ "The capital of China is Beijing.",
252
+ "Gravity is a force that attracts two bodies towards each other. It gives weight to physical objects and is responsible for the movement of planets around the sun.",
253
+ ]
254
+
255
+ pairs = list(zip(queries, documents))
256
+ inputs = process_inputs(pairs, task, max_length-len(suffix_tokens), suffix_tokens)
257
+ scores = compute_logits(model, inputs, sampling_params, true_token, false_token)
258
+ print('scores', scores)
259
+
260
+ destroy_model_parallel()
261
+ ```
262
+
263
+ 📌 **Tip**: We recommend that developers customize the `instruct` according to their specific scenarios, tasks, and languages. Our tests have shown that in most retrieval scenarios, not using an `instruct` on the query side can lead to a drop in retrieval performance by approximately 1% to 5%.
264
+
265
+ ## Evaluation
266
+
267
+ | Model | Param | MTEB-R | CMTEB-R | MMTEB-R | MLDR | MTEB-Code | FollowIR |
268
+ |------------------------------------|--------|---------|---------|---------|--------|-----------|----------|
269
+ | **Qwen3-Embedding-0.6B** | 0.6B | 61.82 | 71.02 | 64.64 | 50.26 | 75.41 | 5.09 |
270
+ | Jina-multilingual-reranker-v2-base | 0.3B | 58.22 | 63.37 | 63.73 | 39.66 | 58.98 | -0.68 |
271
+ | gte-multilingual-reranker-base | 0.3B | 59.51 | 74.08 | 59.44 | 66.33 | 54.18 | -1.64 |
272
+ | BGE-reranker-v2-m3 | 0.6B | 57.03 | 72.16 | 58.36 | 59.51 | 41.38 | -0.01 |
273
+ | **Qwen3-Reranker-0.6B** | 0.6B | 65.80 | 71.31 | 66.36 | 67.28 | 73.42 | 5.41 |
274
+ | **Qwen3-Reranker-4B** | 4B | **69.76** | 75.94 | 72.74 | 69.97 | 81.20 | **14.84** |
275
+ | **Qwen3-Reranker-8B** | 8B | 69.02 | **77.45** | **72.94** | **70.19** | **81.22** | 8.05 |
276
+
277
+ > **Note**:
278
+ > - Evaluation results for reranking models. We use the retrieval subsets of MTEB(eng, v2), MTEB(cmn, v1), MMTEB and MTEB (Code), which are MTEB-R, CMTEB-R, MMTEB-R and MTEB-Code.
279
+ > - All scores are our runs based on the top-100 candidates retrieved by dense embedding model [Qwen3-Embedding-0.6B](https://huggingface.co/Qwen/Qwen3-Embedding-0.6B).
280
+
281
+ ## Citation
282
+ If you find our work helpful, feel free to give us a cite.
283
+
284
+ ```
285
+ @article{qwen3embedding,
286
+ title={Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models},
287
+ author={Zhang, Yanzhao and Li, Mingxin and Long, Dingkun and Zhang, Xin and Lin, Huan and Yang, Baosong and Xie, Pengjun and Yang, An and Liu, Dayiheng and Lin, Junyang and Huang, Fei and Zhou, Jingren},
288
+ journal={arXiv preprint arXiv:2506.05176},
289
+ year={2025}
290
+ }
291
+ ```
qwen3-reranker-0.6b/artifact-manifest.json ADDED
@@ -0,0 +1,94 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "artifactID": "bd81700d5afe9427ab354e4155c77a1cf718c7b253d7c167fd828ca76eb7daf8",
3
+ "contractVersion": 1,
4
+ "coreAICoreVersion": "1.0.0b2",
5
+ "coreAITorchVersion": "0.4.1",
6
+ "files": [
7
+ {
8
+ "byteCount": 14742,
9
+ "path": "SOURCE-README.md",
10
+ "sha256": "5bba8c734f6dd3ae48317b4139317e45a7fce48fc55e15670b23a0dd15492ab6"
11
+ },
12
+ {
13
+ "byteCount": 222,
14
+ "path": "compatibility-marker.json",
15
+ "sha256": "88206a3757a27a746bc47dda42cc00e1a1331e4da6ac5c40eb52cf2685104d34"
16
+ },
17
+ {
18
+ "byteCount": 1356,
19
+ "path": "export-metadata.json",
20
+ "sha256": "e046edbb8576942d0cd2ec3ddb07a4aa8428284c76624619ffdb03f82d8724bd"
21
+ },
22
+ {
23
+ "byteCount": 2412,
24
+ "path": "parity-report.json",
25
+ "sha256": "f6ee2245aadd6d1353600343997fe05ec943ea1c69242ed8c728acc827be0f25"
26
+ },
27
+ {
28
+ "byteCount": 32,
29
+ "path": "qwen3-reranker-0.6b_float16_s512_static.aimodel/main.hash",
30
+ "sha256": "134bec116ce705528d66f47ce9e6d8d3228e28ecc04380c7bd74358bdf9e97d6"
31
+ },
32
+ {
33
+ "byteCount": 1192687632,
34
+ "path": "qwen3-reranker-0.6b_float16_s512_static.aimodel/main.mlirb",
35
+ "sha256": "5978b9117f2a2b9e40bed4135aaad5cce976e8e31f042e228ba0dadff720e714"
36
+ },
37
+ {
38
+ "byteCount": 334,
39
+ "path": "qwen3-reranker-0.6b_float16_s512_static.aimodel/metadata.json",
40
+ "sha256": "8782f6a734c4f215acead4ad6ac2cd215105368a49fdd57f8586fc48befbcf0c"
41
+ },
42
+ {
43
+ "byteCount": 113727,
44
+ "path": "reference.json",
45
+ "sha256": "134e3ab21fd252452fcde142ff1dd4de0c52ddb64f93f4e1820c2babe15745a0"
46
+ },
47
+ {
48
+ "byteCount": 1257,
49
+ "path": "reranker-contract.json",
50
+ "sha256": "e2e83e9605959c4af840284cacb4aa328cade64dcd26d3dbb8a90a5372d442ce"
51
+ },
52
+ {
53
+ "byteCount": 1314,
54
+ "path": "source-revision-receipt.json",
55
+ "sha256": "849145e0dbf4b9665b8623b908f846a1223f713aff98974b8e8fd6af8e194fa8"
56
+ },
57
+ {
58
+ "byteCount": 9838,
59
+ "path": "swift-tokenizer-golden.json",
60
+ "sha256": "975b0852c1198287773194cfbacb7d9595124fe43a481f48171c4067ba3d4917"
61
+ },
62
+ {
63
+ "byteCount": 741,
64
+ "path": "tokenizer/chat_template.jinja",
65
+ "sha256": "6f682162495ec5b39fd9005c01b6aa2a74669379fe967039f1e2cbbe8752369d"
66
+ },
67
+ {
68
+ "byteCount": 11422650,
69
+ "path": "tokenizer/tokenizer.json",
70
+ "sha256": "be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506"
71
+ },
72
+ {
73
+ "byteCount": 376,
74
+ "path": "tokenizer/tokenizer_config.json",
75
+ "sha256": "779410a34a98fb2c6c925a1be6f24d632f1723473161b037f6676eb6c04886c6"
76
+ },
77
+ {
78
+ "byteCount": 579,
79
+ "path": "validation-report.json",
80
+ "sha256": "19070dbaadbc07baacf868f75b4d0c51cf6d4bfb4df9e34f960ee306690f12a0"
81
+ }
82
+ ],
83
+ "modelAssetName": "qwen3-reranker-0.6b_float16_s512_static.aimodel",
84
+ "modelSHA256": "5978b9117f2a2b9e40bed4135aaad5cce976e8e31f042e228ba0dadff720e714",
85
+ "precision": "float16",
86
+ "pythonVersion": "3.12.7",
87
+ "schemaVersion": 1,
88
+ "seed": 0,
89
+ "sequenceLength": 512,
90
+ "sourceModel": "Qwen/Qwen3-Reranker-0.6B",
91
+ "sourceRevision": "e61197ed45024b0ed8a2d74b80b4d909f1255473",
92
+ "torchVersion": "2.11.0",
93
+ "transformersVersion": "5.12.0"
94
+ }
qwen3-reranker-0.6b/compatibility-marker.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "assetVersion": "2.0",
3
+ "coreAICoreVersion": "1.0.0b2",
4
+ "coreAITorchVersion": "0.4.1",
5
+ "minimumMacOSMajorVersion": 27,
6
+ "producer": "coreai-core 1.0.0b2",
7
+ "schemaVersion": 1,
8
+ "validatedOSBuild": "26A5378n"
9
+ }
qwen3-reranker-0.6b/export-metadata.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "assetVersion": "2.0",
3
+ "coreAICoreVersion": "1.0.0b2",
4
+ "coreAITorchVersion": "0.4.1",
5
+ "expectedCoreAICoreVersion": "1.0.0b2",
6
+ "expectedCoreAITorchVersion": "0.4.1",
7
+ "maximumTorchWrapperDelta": 7.82310962677002e-07,
8
+ "minimumMacOSMajorVersion": 27,
9
+ "precision": "float16",
10
+ "producer": "coreai-core 1.0.0b2",
11
+ "promptContractSHA256": "73bfdb454165a637e37f0e33e65fc0ab11fc605c86909cdcc9cd777d6d3a8c7b",
12
+ "pythonVersion": "3.12.7",
13
+ "schemaVersion": 1,
14
+ "seed": 0,
15
+ "sequenceLength": 512,
16
+ "sourceFilesSHA256": {
17
+ "README.md": "5bba8c734f6dd3ae48317b4139317e45a7fce48fc55e15670b23a0dd15492ab6",
18
+ "chat_template.jinja": "6f682162495ec5b39fd9005c01b6aa2a74669379fe967039f1e2cbbe8752369d",
19
+ "config.json": "d479c427a9ca5295218063d4f9aca4f297ab4ac27487cca7af42c84643d51ef0",
20
+ "generation_config.json": "81051cd3f6e77013827148d0b8a6ead93f8ac390d5ab805f849199f0af6a08db",
21
+ "model.safetensors": "27cd75a405b9c1b46b59abfd88aaa209e6fed2a1972cde9b70e7659537c5e65b",
22
+ "tokenizer.json": "aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4",
23
+ "tokenizer_config.json": "253153d0738ceb4c668d2eff957714dd2bea0b56de772a9fdccd96cbf517e6a0"
24
+ },
25
+ "sourceModel": "Qwen/Qwen3-Reranker-0.6B",
26
+ "sourceRevision": "e61197ed45024b0ed8a2d74b80b4d909f1255473",
27
+ "torchVersion": "2.11.0",
28
+ "transformersVersion": "5.12.0"
29
+ }
qwen3-reranker-0.6b/installation-receipt.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "compatibilityMarkerSHA256": "88206a3757a27a746bc47dda42cc00e1a1331e4da6ac5c40eb52cf2685104d34",
3
+ "descriptor": {
4
+ "artifactID": "bd81700d5afe9427ab354e4155c77a1cf718c7b253d7c167fd828ca76eb7daf8",
5
+ "byteCount": 1192687632,
6
+ "contractVersion": 1,
7
+ "manifestSHA256": "5ccfcb4fc6a54a89ee8f3f701ef9e5e18d80b4edd79ae3f4631994b4fecadac1",
8
+ "modelAssetName": "qwen3-reranker-0.6b_float16_s512_static.aimodel",
9
+ "modelSHA256": "5978b9117f2a2b9e40bed4135aaad5cce976e8e31f042e228ba0dadff720e714",
10
+ "schemaVersion": 1,
11
+ "sourceRevision": "e61197ed45024b0ed8a2d74b80b4d909f1255473"
12
+ },
13
+ "installedAt": "2026-07-20T00:01:37.625393+00:00",
14
+ "schemaVersion": 1,
15
+ "validationReportSHA256": "19070dbaadbc07baacf868f75b4d0c51cf6d4bfb4df9e34f960ee306690f12a0"
16
+ }
qwen3-reranker-0.6b/parity-report.json ADDED
@@ -0,0 +1,92 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "artifactID": "bd81700d5afe9427ab354e4155c77a1cf718c7b253d7c167fd828ca76eb7daf8",
3
+ "checks": {
4
+ "compatibilityMetadataValid": true,
5
+ "contractValid": true,
6
+ "operatingSystemValid": true,
7
+ "provenanceValid": true,
8
+ "rankingsPreserved": true,
9
+ "referenceInputsValid": true,
10
+ "scoreRangeValid": true,
11
+ "shapeValid": true,
12
+ "sourceHashesValid": true,
13
+ "swiftTokenizerGoldenValid": true,
14
+ "tokenizerValid": true,
15
+ "truncationValid": true
16
+ },
17
+ "coreAIProbabilities": {
18
+ "beesting_irrelevant": [
19
+ 1,
20
+ 3.6120415e-05
21
+ ],
22
+ "beesting_relevant": [
23
+ 0.018692017,
24
+ 0.9814453
25
+ ],
26
+ "capital_irrelevant": [
27
+ 1,
28
+ 8.285046e-06
29
+ ],
30
+ "capital_relevant": [
31
+ 0.006290436,
32
+ 0.99365234
33
+ ],
34
+ "fuji_irrelevant_ja": [
35
+ 1,
36
+ 8.046627e-06
37
+ ],
38
+ "fuji_relevant_ja": [
39
+ 0.00022339821,
40
+ 1
41
+ ],
42
+ "vat_irrelevant_noisy": [
43
+ 0.9995117,
44
+ 0.000259161
45
+ ],
46
+ "vat_relevant_noisy": [
47
+ 0.9663086,
48
+ 0.033843994
49
+ ]
50
+ },
51
+ "coreAIRankings": {
52
+ "beesting_en": true,
53
+ "capital_en": true,
54
+ "fuji_ja": true,
55
+ "vat_noisy": true
56
+ },
57
+ "maximumCoreAIScoreDelta": 0.0002232193946838379,
58
+ "maximumTorchWrapperDelta": 7.82310962677002e-07,
59
+ "officialScores": {
60
+ "beesting_irrelevant": 3.616731191868894e-05,
61
+ "beesting_relevant": 0.9814824461936951,
62
+ "capital_irrelevant": 8.451959729427472e-06,
63
+ "capital_relevant": 0.9937747716903687,
64
+ "fuji_irrelevant_ja": 8.115292985166889e-06,
65
+ "fuji_relevant_ja": 0.9997767806053162,
66
+ "vat_irrelevant_noisy": 0.0002601594023872167,
67
+ "vat_relevant_noisy": 0.03391358256340027
68
+ },
69
+ "schemaVersion": 1,
70
+ "thresholds": {
71
+ "maximumCoreAIScoreDelta": 0.01,
72
+ "maximumProbabilitySumDeviation": 0.02,
73
+ "maximumTorchWrapperDelta": 0.001
74
+ },
75
+ "validatedAt": "2026-07-20T00:01:36.677715+00:00",
76
+ "wrapperRankings": {
77
+ "beesting_en": true,
78
+ "capital_en": true,
79
+ "fuji_ja": true,
80
+ "vat_noisy": true
81
+ },
82
+ "wrapperScores": {
83
+ "beesting_irrelevant": 3.616727553890087e-05,
84
+ "beesting_relevant": 0.9814824461936951,
85
+ "capital_irrelevant": 8.45198428578442e-06,
86
+ "capital_relevant": 0.9937748908996582,
87
+ "fuji_irrelevant_ja": 8.115192031254992e-06,
88
+ "fuji_relevant_ja": 0.9997767806053162,
89
+ "vat_irrelevant_noisy": 0.0002601571613922715,
90
+ "vat_relevant_noisy": 0.03391280025243759
91
+ }
92
+ }
qwen3-reranker-0.6b/qwen3-reranker-0.6b_float16_s512_static.aimodel/main.hash ADDED
@@ -0,0 +1 @@
 
 
1
+ Yx�*+�@��Z����v��."����� �
qwen3-reranker-0.6b/qwen3-reranker-0.6b_float16_s512_static.aimodel/main.mlirb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5978b9117f2a2b9e40bed4135aaad5cce976e8e31f042e228ba0dadff720e714
3
+ size 1192687632
qwen3-reranker-0.6b/qwen3-reranker-0.6b_float16_s512_static.aimodel/metadata.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "creationDate" : "20260720T000127Z",
3
+ "license" : "Apache-2.0",
4
+ "author" : "Alibaba Qwen",
5
+ "assetVersion" : "2.0",
6
+ "producer" : "coreai-core 1.0.0b2",
7
+ "description" : "Qwen3-Reranker-0.6B, pinned at e61197ed45024b0ed8a2d74b80b4d909f1255473; fp16 yes\/no relevance probabilities on a static 512-token right-padded grid."
8
+ }
qwen3-reranker-0.6b/reference.json ADDED
The diff for this file is too large to render. See raw diff
 
qwen3-reranker-0.6b/reranker-contract.json ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "attentionMask": {
3
+ "name": "attention_mask",
4
+ "scalarType": "int32",
5
+ "shape": [
6
+ 1,
7
+ 512
8
+ ]
9
+ },
10
+ "contractVersion": 1,
11
+ "defaultInstruction": "Given a web search query, retrieve relevant passages that answer the query",
12
+ "inputIDs": {
13
+ "name": "input_ids",
14
+ "scalarType": "int32",
15
+ "shape": [
16
+ 1,
17
+ 512
18
+ ]
19
+ },
20
+ "localNoteDocumentTemplate": "Title: {title}\nRelative path: {relativePath}\nMarkdown:\n{markdown}",
21
+ "noTokenID": 2152,
22
+ "output": {
23
+ "name": "probs",
24
+ "order": [
25
+ "no",
26
+ "yes"
27
+ ],
28
+ "relevanceIndex": 1,
29
+ "scalarType": "float16",
30
+ "shape": [
31
+ 1,
32
+ 2
33
+ ]
34
+ },
35
+ "padTokenID": 151643,
36
+ "paddingSide": "right",
37
+ "pairTemplate": "<Instruct>: {instruction}\n<Query>: {query}\n<Document>: {document}",
38
+ "prefix": "<|im_start|>system\nJudge whether the Document meets the requirements based on the Query and the Instruct provided. Note that the answer can only be \"yes\" or \"no\".<|im_end|>\n<|im_start|>user\n",
39
+ "promptContractSHA256": "73bfdb454165a637e37f0e33e65fc0ab11fc605c86909cdcc9cd777d6d3a8c7b",
40
+ "schemaVersion": 1,
41
+ "suffix": "<|im_end|>\n<|im_start|>assistant\n<think>\n\n</think>\n\n",
42
+ "truncation": "middle",
43
+ "yesTokenID": 9693
44
+ }
qwen3-reranker-0.6b/source-revision-receipt.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "knownSourceSHA256": {
3
+ "model.safetensors": "27cd75a405b9c1b46b59abfd88aaa209e6fed2a1972cde9b70e7659537c5e65b",
4
+ "tokenizer.json": "aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4"
5
+ },
6
+ "repository": "Qwen/Qwen3-Reranker-0.6B",
7
+ "requiredSourcePaths": [
8
+ "README.md",
9
+ "config.json",
10
+ "generation_config.json",
11
+ "model.safetensors",
12
+ "tokenizer.json",
13
+ "tokenizer_config.json",
14
+ "chat_template.jinja"
15
+ ],
16
+ "resolvedPath": "/private/tmp/tickoff-local-notes-rag-reranker-export/source/Qwen3-Reranker-0.6B",
17
+ "revision": "e61197ed45024b0ed8a2d74b80b4d909f1255473",
18
+ "schemaVersion": 1,
19
+ "sourceFilesSHA256": {
20
+ "README.md": "5bba8c734f6dd3ae48317b4139317e45a7fce48fc55e15670b23a0dd15492ab6",
21
+ "chat_template.jinja": "6f682162495ec5b39fd9005c01b6aa2a74669379fe967039f1e2cbbe8752369d",
22
+ "config.json": "d479c427a9ca5295218063d4f9aca4f297ab4ac27487cca7af42c84643d51ef0",
23
+ "generation_config.json": "81051cd3f6e77013827148d0b8a6ead93f8ac390d5ab805f849199f0af6a08db",
24
+ "model.safetensors": "27cd75a405b9c1b46b59abfd88aaa209e6fed2a1972cde9b70e7659537c5e65b",
25
+ "tokenizer.json": "aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4",
26
+ "tokenizer_config.json": "253153d0738ceb4c668d2eff957714dd2bea0b56de772a9fdccd96cbf517e6a0"
27
+ }
28
+ }
qwen3-reranker-0.6b/swift-tokenizer-golden.json ADDED
@@ -0,0 +1,1039 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "attentionMask": [
3
+ 1,
4
+ 1,
5
+ 1,
6
+ 1,
7
+ 1,
8
+ 1,
9
+ 1,
10
+ 1,
11
+ 1,
12
+ 1,
13
+ 1,
14
+ 1,
15
+ 1,
16
+ 1,
17
+ 1,
18
+ 1,
19
+ 1,
20
+ 1,
21
+ 1,
22
+ 1,
23
+ 1,
24
+ 1,
25
+ 1,
26
+ 1,
27
+ 1,
28
+ 1,
29
+ 1,
30
+ 1,
31
+ 1,
32
+ 1,
33
+ 1,
34
+ 1,
35
+ 1,
36
+ 1,
37
+ 1,
38
+ 1,
39
+ 1,
40
+ 1,
41
+ 1,
42
+ 1,
43
+ 1,
44
+ 1,
45
+ 1,
46
+ 1,
47
+ 1,
48
+ 1,
49
+ 1,
50
+ 1,
51
+ 1,
52
+ 1,
53
+ 1,
54
+ 1,
55
+ 1,
56
+ 1,
57
+ 1,
58
+ 1,
59
+ 1,
60
+ 1,
61
+ 1,
62
+ 1,
63
+ 1,
64
+ 1,
65
+ 1,
66
+ 1,
67
+ 1,
68
+ 1,
69
+ 1,
70
+ 1,
71
+ 1,
72
+ 1,
73
+ 1,
74
+ 1,
75
+ 1,
76
+ 1,
77
+ 1,
78
+ 1,
79
+ 1,
80
+ 1,
81
+ 1,
82
+ 1,
83
+ 1,
84
+ 1,
85
+ 1,
86
+ 1,
87
+ 1,
88
+ 1,
89
+ 1,
90
+ 1,
91
+ 1,
92
+ 1,
93
+ 1,
94
+ 1,
95
+ 1,
96
+ 1,
97
+ 1,
98
+ 1,
99
+ 1,
100
+ 1,
101
+ 1,
102
+ 1,
103
+ 1,
104
+ 1,
105
+ 1,
106
+ 1,
107
+ 1,
108
+ 1,
109
+ 1,
110
+ 1,
111
+ 1,
112
+ 0,
113
+ 0,
114
+ 0,
115
+ 0,
116
+ 0,
117
+ 0,
118
+ 0,
119
+ 0,
120
+ 0,
121
+ 0,
122
+ 0,
123
+ 0,
124
+ 0,
125
+ 0,
126
+ 0,
127
+ 0,
128
+ 0,
129
+ 0,
130
+ 0,
131
+ 0,
132
+ 0,
133
+ 0,
134
+ 0,
135
+ 0,
136
+ 0,
137
+ 0,
138
+ 0,
139
+ 0,
140
+ 0,
141
+ 0,
142
+ 0,
143
+ 0,
144
+ 0,
145
+ 0,
146
+ 0,
147
+ 0,
148
+ 0,
149
+ 0,
150
+ 0,
151
+ 0,
152
+ 0,
153
+ 0,
154
+ 0,
155
+ 0,
156
+ 0,
157
+ 0,
158
+ 0,
159
+ 0,
160
+ 0,
161
+ 0,
162
+ 0,
163
+ 0,
164
+ 0,
165
+ 0,
166
+ 0,
167
+ 0,
168
+ 0,
169
+ 0,
170
+ 0,
171
+ 0,
172
+ 0,
173
+ 0,
174
+ 0,
175
+ 0,
176
+ 0,
177
+ 0,
178
+ 0,
179
+ 0,
180
+ 0,
181
+ 0,
182
+ 0,
183
+ 0,
184
+ 0,
185
+ 0,
186
+ 0,
187
+ 0,
188
+ 0,
189
+ 0,
190
+ 0,
191
+ 0,
192
+ 0,
193
+ 0,
194
+ 0,
195
+ 0,
196
+ 0,
197
+ 0,
198
+ 0,
199
+ 0,
200
+ 0,
201
+ 0,
202
+ 0,
203
+ 0,
204
+ 0,
205
+ 0,
206
+ 0,
207
+ 0,
208
+ 0,
209
+ 0,
210
+ 0,
211
+ 0,
212
+ 0,
213
+ 0,
214
+ 0,
215
+ 0,
216
+ 0,
217
+ 0,
218
+ 0,
219
+ 0,
220
+ 0,
221
+ 0,
222
+ 0,
223
+ 0,
224
+ 0,
225
+ 0,
226
+ 0,
227
+ 0,
228
+ 0,
229
+ 0,
230
+ 0,
231
+ 0,
232
+ 0,
233
+ 0,
234
+ 0,
235
+ 0,
236
+ 0,
237
+ 0,
238
+ 0,
239
+ 0,
240
+ 0,
241
+ 0,
242
+ 0,
243
+ 0,
244
+ 0,
245
+ 0,
246
+ 0,
247
+ 0,
248
+ 0,
249
+ 0,
250
+ 0,
251
+ 0,
252
+ 0,
253
+ 0,
254
+ 0,
255
+ 0,
256
+ 0,
257
+ 0,
258
+ 0,
259
+ 0,
260
+ 0,
261
+ 0,
262
+ 0,
263
+ 0,
264
+ 0,
265
+ 0,
266
+ 0,
267
+ 0,
268
+ 0,
269
+ 0,
270
+ 0,
271
+ 0,
272
+ 0,
273
+ 0,
274
+ 0,
275
+ 0,
276
+ 0,
277
+ 0,
278
+ 0,
279
+ 0,
280
+ 0,
281
+ 0,
282
+ 0,
283
+ 0,
284
+ 0,
285
+ 0,
286
+ 0,
287
+ 0,
288
+ 0,
289
+ 0,
290
+ 0,
291
+ 0,
292
+ 0,
293
+ 0,
294
+ 0,
295
+ 0,
296
+ 0,
297
+ 0,
298
+ 0,
299
+ 0,
300
+ 0,
301
+ 0,
302
+ 0,
303
+ 0,
304
+ 0,
305
+ 0,
306
+ 0,
307
+ 0,
308
+ 0,
309
+ 0,
310
+ 0,
311
+ 0,
312
+ 0,
313
+ 0,
314
+ 0,
315
+ 0,
316
+ 0,
317
+ 0,
318
+ 0,
319
+ 0,
320
+ 0,
321
+ 0,
322
+ 0,
323
+ 0,
324
+ 0,
325
+ 0,
326
+ 0,
327
+ 0,
328
+ 0,
329
+ 0,
330
+ 0,
331
+ 0,
332
+ 0,
333
+ 0,
334
+ 0,
335
+ 0,
336
+ 0,
337
+ 0,
338
+ 0,
339
+ 0,
340
+ 0,
341
+ 0,
342
+ 0,
343
+ 0,
344
+ 0,
345
+ 0,
346
+ 0,
347
+ 0,
348
+ 0,
349
+ 0,
350
+ 0,
351
+ 0,
352
+ 0,
353
+ 0,
354
+ 0,
355
+ 0,
356
+ 0,
357
+ 0,
358
+ 0,
359
+ 0,
360
+ 0,
361
+ 0,
362
+ 0,
363
+ 0,
364
+ 0,
365
+ 0,
366
+ 0,
367
+ 0,
368
+ 0,
369
+ 0,
370
+ 0,
371
+ 0,
372
+ 0,
373
+ 0,
374
+ 0,
375
+ 0,
376
+ 0,
377
+ 0,
378
+ 0,
379
+ 0,
380
+ 0,
381
+ 0,
382
+ 0,
383
+ 0,
384
+ 0,
385
+ 0,
386
+ 0,
387
+ 0,
388
+ 0,
389
+ 0,
390
+ 0,
391
+ 0,
392
+ 0,
393
+ 0,
394
+ 0,
395
+ 0,
396
+ 0,
397
+ 0,
398
+ 0,
399
+ 0,
400
+ 0,
401
+ 0,
402
+ 0,
403
+ 0,
404
+ 0,
405
+ 0,
406
+ 0,
407
+ 0,
408
+ 0,
409
+ 0,
410
+ 0,
411
+ 0,
412
+ 0,
413
+ 0,
414
+ 0,
415
+ 0,
416
+ 0,
417
+ 0,
418
+ 0,
419
+ 0,
420
+ 0,
421
+ 0,
422
+ 0,
423
+ 0,
424
+ 0,
425
+ 0,
426
+ 0,
427
+ 0,
428
+ 0,
429
+ 0,
430
+ 0,
431
+ 0,
432
+ 0,
433
+ 0,
434
+ 0,
435
+ 0,
436
+ 0,
437
+ 0,
438
+ 0,
439
+ 0,
440
+ 0,
441
+ 0,
442
+ 0,
443
+ 0,
444
+ 0,
445
+ 0,
446
+ 0,
447
+ 0,
448
+ 0,
449
+ 0,
450
+ 0,
451
+ 0,
452
+ 0,
453
+ 0,
454
+ 0,
455
+ 0,
456
+ 0,
457
+ 0,
458
+ 0,
459
+ 0,
460
+ 0,
461
+ 0,
462
+ 0,
463
+ 0,
464
+ 0,
465
+ 0,
466
+ 0,
467
+ 0,
468
+ 0,
469
+ 0,
470
+ 0,
471
+ 0,
472
+ 0,
473
+ 0,
474
+ 0,
475
+ 0,
476
+ 0,
477
+ 0,
478
+ 0,
479
+ 0,
480
+ 0,
481
+ 0,
482
+ 0,
483
+ 0,
484
+ 0,
485
+ 0,
486
+ 0,
487
+ 0,
488
+ 0,
489
+ 0,
490
+ 0,
491
+ 0,
492
+ 0,
493
+ 0,
494
+ 0,
495
+ 0,
496
+ 0,
497
+ 0,
498
+ 0,
499
+ 0,
500
+ 0,
501
+ 0,
502
+ 0,
503
+ 0,
504
+ 0,
505
+ 0,
506
+ 0,
507
+ 0,
508
+ 0,
509
+ 0,
510
+ 0,
511
+ 0,
512
+ 0,
513
+ 0,
514
+ 0
515
+ ],
516
+ "contractVersion": 1,
517
+ "inputIDs": [
518
+ 151644,
519
+ 8948,
520
+ 198,
521
+ 60256,
522
+ 3425,
523
+ 279,
524
+ 11789,
525
+ 20027,
526
+ 279,
527
+ 8502,
528
+ 3118,
529
+ 389,
530
+ 279,
531
+ 11361,
532
+ 323,
533
+ 279,
534
+ 758,
535
+ 1235,
536
+ 3897,
537
+ 13,
538
+ 7036,
539
+ 429,
540
+ 279,
541
+ 4226,
542
+ 646,
543
+ 1172,
544
+ 387,
545
+ 330,
546
+ 9693,
547
+ 1,
548
+ 476,
549
+ 330,
550
+ 2152,
551
+ 3263,
552
+ 151645,
553
+ 198,
554
+ 151644,
555
+ 872,
556
+ 198,
557
+ 27,
558
+ 641,
559
+ 1235,
560
+ 26818,
561
+ 16246,
562
+ 264,
563
+ 3482,
564
+ 2711,
565
+ 3239,
566
+ 11,
567
+ 17179,
568
+ 9760,
569
+ 46769,
570
+ 429,
571
+ 4226,
572
+ 279,
573
+ 3239,
574
+ 198,
575
+ 27,
576
+ 2859,
577
+ 26818,
578
+ 2585,
579
+ 525,
580
+ 47592,
581
+ 1372,
582
+ 374,
583
+ 5267,
584
+ 75692,
585
+ 26818,
586
+ 10869,
587
+ 25,
588
+ 15098,
589
+ 15263,
590
+ 220,
591
+ 16,
592
+ 21,
593
+ 198,
594
+ 28442,
595
+ 1815,
596
+ 25,
597
+ 15098,
598
+ 15263,
599
+ 220,
600
+ 16,
601
+ 21,
602
+ 21324,
603
+ 198,
604
+ 68005,
605
+ 510,
606
+ 53,
607
+ 828,
608
+ 18865,
609
+ 19,
610
+ 19,
611
+ 20,
612
+ 23,
613
+ 24,
614
+ 17,
615
+ 16,
616
+ 16,
617
+ 18,
618
+ 151645,
619
+ 198,
620
+ 151644,
621
+ 77091,
622
+ 198,
623
+ 151667,
624
+ 271,
625
+ 151668,
626
+ 271,
627
+ 151643,
628
+ 151643,
629
+ 151643,
630
+ 151643,
631
+ 151643,
632
+ 151643,
633
+ 151643,
634
+ 151643,
635
+ 151643,
636
+ 151643,
637
+ 151643,
638
+ 151643,
639
+ 151643,
640
+ 151643,
641
+ 151643,
642
+ 151643,
643
+ 151643,
644
+ 151643,
645
+ 151643,
646
+ 151643,
647
+ 151643,
648
+ 151643,
649
+ 151643,
650
+ 151643,
651
+ 151643,
652
+ 151643,
653
+ 151643,
654
+ 151643,
655
+ 151643,
656
+ 151643,
657
+ 151643,
658
+ 151643,
659
+ 151643,
660
+ 151643,
661
+ 151643,
662
+ 151643,
663
+ 151643,
664
+ 151643,
665
+ 151643,
666
+ 151643,
667
+ 151643,
668
+ 151643,
669
+ 151643,
670
+ 151643,
671
+ 151643,
672
+ 151643,
673
+ 151643,
674
+ 151643,
675
+ 151643,
676
+ 151643,
677
+ 151643,
678
+ 151643,
679
+ 151643,
680
+ 151643,
681
+ 151643,
682
+ 151643,
683
+ 151643,
684
+ 151643,
685
+ 151643,
686
+ 151643,
687
+ 151643,
688
+ 151643,
689
+ 151643,
690
+ 151643,
691
+ 151643,
692
+ 151643,
693
+ 151643,
694
+ 151643,
695
+ 151643,
696
+ 151643,
697
+ 151643,
698
+ 151643,
699
+ 151643,
700
+ 151643,
701
+ 151643,
702
+ 151643,
703
+ 151643,
704
+ 151643,
705
+ 151643,
706
+ 151643,
707
+ 151643,
708
+ 151643,
709
+ 151643,
710
+ 151643,
711
+ 151643,
712
+ 151643,
713
+ 151643,
714
+ 151643,
715
+ 151643,
716
+ 151643,
717
+ 151643,
718
+ 151643,
719
+ 151643,
720
+ 151643,
721
+ 151643,
722
+ 151643,
723
+ 151643,
724
+ 151643,
725
+ 151643,
726
+ 151643,
727
+ 151643,
728
+ 151643,
729
+ 151643,
730
+ 151643,
731
+ 151643,
732
+ 151643,
733
+ 151643,
734
+ 151643,
735
+ 151643,
736
+ 151643,
737
+ 151643,
738
+ 151643,
739
+ 151643,
740
+ 151643,
741
+ 151643,
742
+ 151643,
743
+ 151643,
744
+ 151643,
745
+ 151643,
746
+ 151643,
747
+ 151643,
748
+ 151643,
749
+ 151643,
750
+ 151643,
751
+ 151643,
752
+ 151643,
753
+ 151643,
754
+ 151643,
755
+ 151643,
756
+ 151643,
757
+ 151643,
758
+ 151643,
759
+ 151643,
760
+ 151643,
761
+ 151643,
762
+ 151643,
763
+ 151643,
764
+ 151643,
765
+ 151643,
766
+ 151643,
767
+ 151643,
768
+ 151643,
769
+ 151643,
770
+ 151643,
771
+ 151643,
772
+ 151643,
773
+ 151643,
774
+ 151643,
775
+ 151643,
776
+ 151643,
777
+ 151643,
778
+ 151643,
779
+ 151643,
780
+ 151643,
781
+ 151643,
782
+ 151643,
783
+ 151643,
784
+ 151643,
785
+ 151643,
786
+ 151643,
787
+ 151643,
788
+ 151643,
789
+ 151643,
790
+ 151643,
791
+ 151643,
792
+ 151643,
793
+ 151643,
794
+ 151643,
795
+ 151643,
796
+ 151643,
797
+ 151643,
798
+ 151643,
799
+ 151643,
800
+ 151643,
801
+ 151643,
802
+ 151643,
803
+ 151643,
804
+ 151643,
805
+ 151643,
806
+ 151643,
807
+ 151643,
808
+ 151643,
809
+ 151643,
810
+ 151643,
811
+ 151643,
812
+ 151643,
813
+ 151643,
814
+ 151643,
815
+ 151643,
816
+ 151643,
817
+ 151643,
818
+ 151643,
819
+ 151643,
820
+ 151643,
821
+ 151643,
822
+ 151643,
823
+ 151643,
824
+ 151643,
825
+ 151643,
826
+ 151643,
827
+ 151643,
828
+ 151643,
829
+ 151643,
830
+ 151643,
831
+ 151643,
832
+ 151643,
833
+ 151643,
834
+ 151643,
835
+ 151643,
836
+ 151643,
837
+ 151643,
838
+ 151643,
839
+ 151643,
840
+ 151643,
841
+ 151643,
842
+ 151643,
843
+ 151643,
844
+ 151643,
845
+ 151643,
846
+ 151643,
847
+ 151643,
848
+ 151643,
849
+ 151643,
850
+ 151643,
851
+ 151643,
852
+ 151643,
853
+ 151643,
854
+ 151643,
855
+ 151643,
856
+ 151643,
857
+ 151643,
858
+ 151643,
859
+ 151643,
860
+ 151643,
861
+ 151643,
862
+ 151643,
863
+ 151643,
864
+ 151643,
865
+ 151643,
866
+ 151643,
867
+ 151643,
868
+ 151643,
869
+ 151643,
870
+ 151643,
871
+ 151643,
872
+ 151643,
873
+ 151643,
874
+ 151643,
875
+ 151643,
876
+ 151643,
877
+ 151643,
878
+ 151643,
879
+ 151643,
880
+ 151643,
881
+ 151643,
882
+ 151643,
883
+ 151643,
884
+ 151643,
885
+ 151643,
886
+ 151643,
887
+ 151643,
888
+ 151643,
889
+ 151643,
890
+ 151643,
891
+ 151643,
892
+ 151643,
893
+ 151643,
894
+ 151643,
895
+ 151643,
896
+ 151643,
897
+ 151643,
898
+ 151643,
899
+ 151643,
900
+ 151643,
901
+ 151643,
902
+ 151643,
903
+ 151643,
904
+ 151643,
905
+ 151643,
906
+ 151643,
907
+ 151643,
908
+ 151643,
909
+ 151643,
910
+ 151643,
911
+ 151643,
912
+ 151643,
913
+ 151643,
914
+ 151643,
915
+ 151643,
916
+ 151643,
917
+ 151643,
918
+ 151643,
919
+ 151643,
920
+ 151643,
921
+ 151643,
922
+ 151643,
923
+ 151643,
924
+ 151643,
925
+ 151643,
926
+ 151643,
927
+ 151643,
928
+ 151643,
929
+ 151643,
930
+ 151643,
931
+ 151643,
932
+ 151643,
933
+ 151643,
934
+ 151643,
935
+ 151643,
936
+ 151643,
937
+ 151643,
938
+ 151643,
939
+ 151643,
940
+ 151643,
941
+ 151643,
942
+ 151643,
943
+ 151643,
944
+ 151643,
945
+ 151643,
946
+ 151643,
947
+ 151643,
948
+ 151643,
949
+ 151643,
950
+ 151643,
951
+ 151643,
952
+ 151643,
953
+ 151643,
954
+ 151643,
955
+ 151643,
956
+ 151643,
957
+ 151643,
958
+ 151643,
959
+ 151643,
960
+ 151643,
961
+ 151643,
962
+ 151643,
963
+ 151643,
964
+ 151643,
965
+ 151643,
966
+ 151643,
967
+ 151643,
968
+ 151643,
969
+ 151643,
970
+ 151643,
971
+ 151643,
972
+ 151643,
973
+ 151643,
974
+ 151643,
975
+ 151643,
976
+ 151643,
977
+ 151643,
978
+ 151643,
979
+ 151643,
980
+ 151643,
981
+ 151643,
982
+ 151643,
983
+ 151643,
984
+ 151643,
985
+ 151643,
986
+ 151643,
987
+ 151643,
988
+ 151643,
989
+ 151643,
990
+ 151643,
991
+ 151643,
992
+ 151643,
993
+ 151643,
994
+ 151643,
995
+ 151643,
996
+ 151643,
997
+ 151643,
998
+ 151643,
999
+ 151643,
1000
+ 151643,
1001
+ 151643,
1002
+ 151643,
1003
+ 151643,
1004
+ 151643,
1005
+ 151643,
1006
+ 151643,
1007
+ 151643,
1008
+ 151643,
1009
+ 151643,
1010
+ 151643,
1011
+ 151643,
1012
+ 151643,
1013
+ 151643,
1014
+ 151643,
1015
+ 151643,
1016
+ 151643,
1017
+ 151643,
1018
+ 151643,
1019
+ 151643,
1020
+ 151643,
1021
+ 151643,
1022
+ 151643,
1023
+ 151643,
1024
+ 151643,
1025
+ 151643,
1026
+ 151643,
1027
+ 151643,
1028
+ 151643,
1029
+ 151643
1030
+ ],
1031
+ "markdown": "VAT GB445892113",
1032
+ "promptContractSHA256": "73bfdb454165a637e37f0e33e65fc0ab11fc605c86909cdcc9cd777d6d3a8c7b",
1033
+ "query": "How are VAT number is?",
1034
+ "realTokenCount": 109,
1035
+ "relativePath": "Untitled 16.md",
1036
+ "schemaVersion": 1,
1037
+ "title": "Untitled 16",
1038
+ "wasTruncated": false
1039
+ }
qwen3-reranker-0.6b/tokenizer/chat_template.jinja ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- set instruction = messages | selectattr("role", "eq", "system") | map(attribute="content") | first | default("Given a web search query, retrieve relevant passages that answer the query") -%}
2
+ {%- set query_text = messages | selectattr("role", "eq", "query") | map(attribute="content") | first -%}
3
+ {%- set document_text = messages | selectattr("role", "eq", "document") | map(attribute="content") | first -%}
4
+ <|im_start|>system
5
+ 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".<|im_end|>
6
+ <|im_start|>user
7
+ <Instruct>: {{ instruction }}
8
+ <Query>: {{ query_text }}
9
+ <Document>: {{ document_text }}<|im_end|>
10
+ <|im_start|>assistant
11
+ <think>
12
+
13
+ </think>
14
+
15
+
qwen3-reranker-0.6b/tokenizer/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
3
+ size 11422650
qwen3-reranker-0.6b/tokenizer/tokenizer_config.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "backend": "tokenizers",
4
+ "bos_token": null,
5
+ "clean_up_tokenization_spaces": false,
6
+ "eos_token": "<|im_end|>",
7
+ "errors": "replace",
8
+ "is_local": true,
9
+ "local_files_only": true,
10
+ "model_max_length": 131072,
11
+ "pad_token": "<|endoftext|>",
12
+ "split_special_tokens": false,
13
+ "tokenizer_class": "Qwen2Tokenizer",
14
+ "unk_token": null
15
+ }
qwen3-reranker-0.6b/validation-report.json ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "artifactID": "bd81700d5afe9427ab354e4155c77a1cf718c7b253d7c167fd828ca76eb7daf8",
3
+ "hardwareModel": "Mac15,8",
4
+ "inputShape": [
5
+ 1,
6
+ 512
7
+ ],
8
+ "maximumCoreAIScoreDelta": 0.0002232193946838379,
9
+ "modelSHA256": "5978b9117f2a2b9e40bed4135aaad5cce976e8e31f042e228ba0dadff720e714",
10
+ "outputScalarType": "float16",
11
+ "outputShape": [
12
+ 1,
13
+ 2
14
+ ],
15
+ "passed": true,
16
+ "rankingsPreserved": true,
17
+ "schemaVersion": 1,
18
+ "scoreRangeValid": true,
19
+ "validatedAt": "2026-07-20T00:01:36.677715+00:00",
20
+ "validatedOSBuild": "26A5378n",
21
+ "validatedOSVersion": "27.0.0"
22
+ }