pipeline_tag	tags	model-index
sentence-similarity	mteb sentence-transformers feature-extraction sentence-similarity	name results acge_text_embedding task dataset metrics type STS type name config split revision C-MTEB/AFQMC MTEB AFQMC default validation b44c3b011063adb25877c13823db83bb193913c4 type value cos_sim_pearson 54.03434872650919 type value cos_sim_spearman 58.80730796688325 type value euclidean_pearson 57.47231387497989 type value euclidean_spearman 58.80775026351807 type value manhattan_pearson 57.46332720141574 type value manhattan_spearman 58.80196022940078 task dataset metrics type STS type name config split revision C-MTEB/ATEC MTEB ATEC default test 0f319b1142f28d00e055a6770f3f726ae9b7d865 type value cos_sim_pearson 53.52621290548175 type value cos_sim_spearman 57.945227768312144 type value euclidean_pearson 61.17041394151802 type value euclidean_spearman 57.94553287835657 type value manhattan_pearson 61.168327500057885 type value manhattan_spearman 57.94477516925043 task dataset metrics type Classification type name config split revision mteb/amazon_reviews_multi MTEB AmazonReviewsClassification (zh) zh test 1399c76144fd37290681b995c656ef9b2e06e26d type value accuracy 48.538000000000004 type value f1 46.59920995594044 task dataset metrics type STS type name config split revision C-MTEB/BQ MTEB BQ default test e3dda5e115e487b39ec7e618c0c6a29137052a55 type value cos_sim_pearson 68.27529991817154 type value cos_sim_spearman 70.37095914176643 type value euclidean_pearson 69.42690712802727 type value euclidean_spearman 70.37017971889912 type value manhattan_pearson 69.40264877917839 type value manhattan_spearman 70.34786744049524 task dataset metrics type Clustering type name config split revision C-MTEB/CLSClusteringP2P MTEB CLSClusteringP2P default test 4b6227591c6c1a73bc76b1055f3b7f3588e72476 type value v_measure 47.08027536192709 task dataset metrics type Clustering type name config split revision C-MTEB/CLSClusteringS2S MTEB CLSClusteringS2S default test e458b3f5414b62b7f9f83499ac1f5497ae2e869f type value v_measure 44.0526024940363 task dataset metrics type Reranking type name config split revision C-MTEB/CMedQAv1-reranking MTEB CMedQAv1 default test 8d7f1e942507dac42dc58017c1a001c3717da7df type value map 88.65974993133156 type value mrr 90.64761904761905 task dataset metrics type Reranking type name config split revision C-MTEB/CMedQAv2-reranking MTEB CMedQAv2 default test 23d186750531a14a0357ca22cd92d712fd512ea0 type value map 88.90396838907245 type value mrr 90.90932539682541 task dataset metrics type Retrieval type name config split revision C-MTEB/CmedqaRetrieval MTEB CmedqaRetrieval default dev cd540c506dae1cf9e9a59c3e06f42030d54e7301 type value map_at_1 26.875 type value map_at_10 39.995999999999995 type value map_at_100 41.899 type value map_at_1000 42.0 type value map_at_3 35.414 type value map_at_5 38.019 type value mrr_at_1 40.635 type value mrr_at_10 48.827 type value mrr_at_100 49.805 type value mrr_at_1000 49.845 type value mrr_at_3 46.145 type value mrr_at_5 47.693999999999996 type value ndcg_at_1 40.635 type value ndcg_at_10 46.78 type value ndcg_at_100 53.986999999999995 type value ndcg_at_1000 55.684 type value ndcg_at_3 41.018 type value ndcg_at_5 43.559 type value precision_at_1 40.635 type value precision_at_10 10.427999999999999 type value precision_at_100 1.625 type value precision_at_1000 0.184 type value precision_at_3 23.139000000000003 type value precision_at_5 17.004 type value recall_at_1 26.875 type value recall_at_10 57.887 type value recall_at_100 87.408 type value recall_at_1000 98.721 type value recall_at_3 40.812 type value recall_at_5 48.397 task dataset metrics type PairClassification type name config split revision C-MTEB/CMNLI MTEB Cmnli default validation 41bc36f332156f7adc9e38f53777c959b2ae9766 type value cos_sim_accuracy 83.43956704750451 type value cos_sim_ap 90.49172854352659 type value cos_sim_f1 84.28475486903963 type value cos_sim_precision 80.84603822203135 type value cos_sim_recall 88.02899228431144 type value dot_accuracy 83.43956704750451 type value dot_ap 90.46317132695233 type value dot_f1 84.28794294628929 type value dot_precision 80.51948051948052 type value dot_recall 88.4264671498714 type value euclidean_accuracy 83.43956704750451 type value euclidean_ap 90.49171785256486 type value euclidean_f1 84.28235820561584 type value euclidean_precision 80.8022308022308 type value euclidean_recall 88.07575403320084 type value manhattan_accuracy 83.55983162958509 type value manhattan_ap 90.48046779812815 type value manhattan_f1 84.45354259069714 type value manhattan_precision 82.21877767936226 type value manhattan_recall 86.81318681318682 type value max_accuracy 83.55983162958509 type value max_ap 90.49172854352659 type value max_f1 84.45354259069714 task dataset metrics type Retrieval type name config split revision C-MTEB/CovidRetrieval MTEB CovidRetrieval default dev 1271c7809071a13532e05f25fb53511ffce77117 type value map_at_1 68.54599999999999 type value map_at_10 77.62400000000001 type value map_at_100 77.886 type value map_at_1000 77.89 type value map_at_3 75.966 type value map_at_5 76.995 type value mrr_at_1 68.915 type value mrr_at_10 77.703 type value mrr_at_100 77.958 type value mrr_at_1000 77.962 type value mrr_at_3 76.08 type value mrr_at_5 77.118 type value ndcg_at_1 68.809 type value ndcg_at_10 81.563 type value ndcg_at_100 82.758 type value ndcg_at_1000 82.864 type value ndcg_at_3 78.29 type value ndcg_at_5 80.113 type value precision_at_1 68.809 type value precision_at_10 9.463000000000001 type value precision_at_100 1.001 type value precision_at_1000 0.101 type value precision_at_3 28.486 type value precision_at_5 18.019 type value recall_at_1 68.54599999999999 type value recall_at_10 93.625 type value recall_at_100 99.05199999999999 type value recall_at_1000 99.895 type value recall_at_3 84.879 type value recall_at_5 89.252 task dataset metrics type Retrieval type name config split revision C-MTEB/DuRetrieval MTEB DuRetrieval default dev a1a333e290fe30b10f3f56498e3a0d911a693ced type value map_at_1 25.653 type value map_at_10 79.105 type value map_at_100 81.902 type value map_at_1000 81.947 type value map_at_3 54.54599999999999 type value map_at_5 69.226 type value mrr_at_1 89.35 type value mrr_at_10 92.69 type value mrr_at_100 92.77 type value mrr_at_1000 92.774 type value mrr_at_3 92.425 type value mrr_at_5 92.575 type value ndcg_at_1 89.35 type value ndcg_at_10 86.55199999999999 type value ndcg_at_100 89.35300000000001 type value ndcg_at_1000 89.782 type value ndcg_at_3 85.392 type value ndcg_at_5 84.5 type value precision_at_1 89.35 type value precision_at_10 41.589999999999996 type value precision_at_100 4.781 type value precision_at_1000 0.488 type value precision_at_3 76.683 type value precision_at_5 65.06 type value recall_at_1 25.653 type value recall_at_10 87.64999999999999 type value recall_at_100 96.858 type value recall_at_1000 99.13300000000001 type value recall_at_3 56.869 type value recall_at_5 74.024 task dataset metrics type Retrieval type name config split revision C-MTEB/EcomRetrieval MTEB EcomRetrieval default dev 687de13dc7294d6fd9be10c6945f9e8fec8166b9 type value map_at_1 52.1 type value map_at_10 62.629999999999995 type value map_at_100 63.117000000000004 type value map_at_1000 63.134 type value map_at_3 60.267 type value map_at_5 61.777 type value mrr_at_1 52.1 type value mrr_at_10 62.629999999999995 type value mrr_at_100 63.117000000000004 type value mrr_at_1000 63.134 type value mrr_at_3 60.267 type value mrr_at_5 61.777 type value ndcg_at_1 52.1 type value ndcg_at_10 67.596 type value ndcg_at_100 69.95 type value ndcg_at_1000 70.33500000000001 type value ndcg_at_3 62.82600000000001 type value ndcg_at_5 65.546 type value precision_at_1 52.1 type value precision_at_10 8.309999999999999 type value precision_at_100 0.941 type value precision_at_1000 0.097 type value precision_at_3 23.400000000000002 type value precision_at_5 15.36 type value recall_at_1 52.1 type value recall_at_10 83.1 type value recall_at_100 94.1 type value recall_at_1000 97.0 type value recall_at_3 70.19999999999999 type value recall_at_5 76.8 task dataset metrics type Classification type name config split revision C-MTEB/IFlyTek-classification MTEB IFlyTek default validation 421605374b29664c5fc098418fe20ada9bd55f8a type value accuracy 51.773759138130046 type value f1 40.341407912920054 task dataset metrics type Classification type name config split revision C-MTEB/JDReview-classification MTEB JDReview default test b7c64bd89eb87f8ded463478346f76731f07bf8b type value accuracy 86.69793621013133 type value ap 55.46718958939327 type value f1 81.48228915952436 task dataset metrics type STS type name config split revision C-MTEB/LCQMC MTEB LCQMC default test 17f9b096f80380fce5ed12a9be8be7784b337daf type value cos_sim_pearson 71.1397780205448 type value cos_sim_spearman 78.17368193033309 type value euclidean_pearson 77.4849177602368 type value euclidean_spearman 78.17369079663212 type value manhattan_pearson 77.47344305182406 type value manhattan_spearman 78.16454335155387 task dataset metrics type Reranking type name config split revision C-MTEB/Mmarco-reranking MTEB MMarcoReranking default dev 8e0c766dbe9e16e1d221116a3f36795fbade07f6 type value map 27.76160559006673 type value mrr 28.02420634920635 task dataset metrics type Retrieval type name config split revision C-MTEB/MMarcoRetrieval MTEB MMarcoRetrieval default dev 539bbde593d947e2a124ba72651aafc09eb33fc2 type value map_at_1 65.661 type value map_at_10 74.752 type value map_at_100 75.091 type value map_at_1000 75.104 type value map_at_3 72.997 type value map_at_5 74.119 type value mrr_at_1 67.923 type value mrr_at_10 75.376 type value mrr_at_100 75.673 type value mrr_at_1000 75.685 type value mrr_at_3 73.856 type value mrr_at_5 74.82799999999999 type value ndcg_at_1 67.923 type value ndcg_at_10 78.424 type value ndcg_at_100 79.95100000000001 type value ndcg_at_1000 80.265 type value ndcg_at_3 75.101 type value ndcg_at_5 76.992 type value precision_at_1 67.923 type value precision_at_10 9.474 type value precision_at_100 1.023 type value precision_at_1000 0.105 type value precision_at_3 28.319 type value precision_at_5 17.986 type value recall_at_1 65.661 type value recall_at_10 89.09899999999999 type value recall_at_100 96.023 type value recall_at_1000 98.455 type value recall_at_3 80.314 type value recall_at_5 84.81 task dataset metrics type Classification type name config split revision mteb/amazon_massive_intent MTEB MassiveIntentClassification (zh-CN) zh-CN test 31efe3c427b0bae9c22cbb560b8f15491cc6bed7 type value accuracy 75.86751849361131 type value f1 73.04918450508 task dataset metrics type Classification type name config split revision mteb/amazon_massive_scenario MTEB MassiveScenarioClassification (zh-CN) zh-CN test 7d571f92784cd94a019292a1f45445077d0ef634 type value accuracy 78.4364492266308 type value f1 78.120686034844 task dataset metrics type Retrieval type name config split revision C-MTEB/MedicalRetrieval MTEB MedicalRetrieval default dev 2039188fb5800a9803ba5048df7b76e6fb151fc6 type value map_at_1 55.00000000000001 type value map_at_10 61.06399999999999 type value map_at_100 61.622 type value map_at_1000 61.663000000000004 type value map_at_3 59.583 type value map_at_5 60.373 type value mrr_at_1 55.2 type value mrr_at_10 61.168 type value mrr_at_100 61.726000000000006 type value mrr_at_1000 61.767 type value mrr_at_3 59.683 type value mrr_at_5 60.492999999999995 type value ndcg_at_1 55.00000000000001 type value ndcg_at_10 64.098 type value ndcg_at_100 67.05 type value ndcg_at_1000 68.262 type value ndcg_at_3 61.00600000000001 type value ndcg_at_5 62.439 type value precision_at_1 55.00000000000001 type value precision_at_10 7.37 type value precision_at_100 0.881 type value precision_at_1000 0.098 type value precision_at_3 21.7 type value precision_at_5 13.719999999999999 type value recall_at_1 55.00000000000001 type value recall_at_10 73.7 type value recall_at_100 88.1 type value recall_at_1000 97.8 type value recall_at_3 65.10000000000001 type value recall_at_5 68.60000000000001 task dataset metrics type Classification type name config split revision C-MTEB/MultilingualSentiment-classification MTEB MultilingualSentiment default validation 46958b007a63fdbf239b7672c25d0bea67b5ea1a type value accuracy 77.52666666666667 type value f1 77.49784731367215 task dataset metrics type PairClassification type name config split revision C-MTEB/OCNLI MTEB Ocnli default validation 66e76a618a34d6d565d5538088562851e6daa7ec type value cos_sim_accuracy 81.10449377368705 type value cos_sim_ap 85.17742765935606 type value cos_sim_f1 83.00094966761633 type value cos_sim_precision 75.40983606557377 type value cos_sim_recall 92.29144667370645 type value dot_accuracy 81.10449377368705 type value dot_ap 85.17143850809614 type value dot_f1 83.01707779886148 type value dot_precision 75.36606373815677 type value dot_recall 92.39704329461456 type value euclidean_accuracy 81.10449377368705 type value euclidean_ap 85.17856775343333 type value euclidean_f1 83.00094966761633 type value euclidean_precision 75.40983606557377 type value euclidean_recall 92.29144667370645 type value manhattan_accuracy 81.05035192203573 type value manhattan_ap 85.14464459395809 type value manhattan_f1 82.96155671570953 type value manhattan_precision 75.3448275862069 type value manhattan_recall 92.29144667370645 type value max_accuracy 81.10449377368705 type value max_ap 85.17856775343333 type value max_f1 83.01707779886148 task dataset metrics type Classification type name config split revision C-MTEB/OnlineShopping-classification MTEB OnlineShopping default test e610f2ebd179a8fda30ae534c3878750a96db120 type value accuracy 93.71000000000001 type value ap 91.83202232349356 type value f1 93.69900560334331 task dataset metrics type STS type name config split revision C-MTEB/PAWSX MTEB PAWSX default test 9c6a90e430ac22b5779fb019a23e820b11a8b5e1 type value cos_sim_pearson 39.175047651512415 type value cos_sim_spearman 45.51434675777896 type value euclidean_pearson 44.864110004132286 type value euclidean_spearman 45.516433048896076 type value manhattan_pearson 44.87153627706517 type value manhattan_spearman 45.52862617925012 task dataset metrics type STS type name config split revision C-MTEB/QBQTC MTEB QBQTC default test 790b0510dc52b1553e8c49f3d2afb48c0e5c48b7 type value cos_sim_pearson 34.249579701429084 type value cos_sim_spearman 37.30903127368978 type value euclidean_pearson 35.129438425253355 type value euclidean_spearman 37.308544018709085 type value manhattan_pearson 35.08936153503652 type value manhattan_spearman 37.25582901077839 task dataset metrics type STS type name config split revision mteb/sts22-crosslingual-sts MTEB STS22 (zh) zh test eea2b4fe26a775864c896887d910b76a8098ad3f type value cos_sim_pearson 61.29309637460004 type value cos_sim_spearman 65.85136090376717 type value euclidean_pearson 64.04783990953557 type value euclidean_spearman 65.85036859610366 type value manhattan_pearson 63.995852552712186 type value manhattan_spearman 65.86508416749417 task dataset metrics type STS type name config split revision C-MTEB/STSB MTEB STSB default test 0cde68302b3541bb8b3c340dc0644b0b745b3dc0 type value cos_sim_pearson 81.5595940455587 type value cos_sim_spearman 82.72654634579749 type value euclidean_pearson 82.4892721061365 type value euclidean_spearman 82.72678504228253 type value manhattan_pearson 82.4770861422454 type value manhattan_spearman 82.71137469783162 task dataset metrics type Reranking type name config split revision C-MTEB/T2Reranking MTEB T2Reranking default dev 76631901a18387f85eaa53e5450019b87ad58ef9 type value map 66.6159547610527 type value mrr 76.35739406347057 task dataset metrics type Retrieval type name config split revision C-MTEB/T2Retrieval MTEB T2Retrieval default dev 8731a845f1bf500a4f111cf1070785c793d10e64 type value map_at_1 27.878999999999998 type value map_at_10 77.517 type value map_at_100 81.139 type value map_at_1000 81.204 type value map_at_3 54.728 type value map_at_5 67.128 type value mrr_at_1 90.509 type value mrr_at_10 92.964 type value mrr_at_100 93.045 type value mrr_at_1000 93.048 type value mrr_at_3 92.551 type value mrr_at_5 92.81099999999999 type value ndcg_at_1 90.509 type value ndcg_at_10 85.075 type value ndcg_at_100 88.656 type value ndcg_at_1000 89.25699999999999 type value ndcg_at_3 86.58200000000001 type value ndcg_at_5 85.138 type value precision_at_1 90.509 type value precision_at_10 42.05 type value precision_at_100 5.013999999999999 type value precision_at_1000 0.516 type value precision_at_3 75.551 type value precision_at_5 63.239999999999995 type value recall_at_1 27.878999999999998 type value recall_at_10 83.941 type value recall_at_100 95.568 type value recall_at_1000 98.55000000000001 type value recall_at_3 56.374 type value recall_at_5 70.435 task dataset metrics type Classification type name config split revision C-MTEB/TNews-classification MTEB TNews default validation 317f262bf1e6126357bbe89e875451e4b0938fe4 type value accuracy 53.687 type value f1 51.86911933364655 task dataset metrics type Clustering type name config split revision C-MTEB/ThuNewsClusteringP2P MTEB ThuNewsClusteringP2P default test 5798586b105c0434e4f0fe5e767abe619442cf93 type value v_measure 74.65887489872564 task dataset metrics type Clustering type name config split revision C-MTEB/ThuNewsClusteringS2S MTEB ThuNewsClusteringS2S default test 8a8b2caeda43f39e13c4bc5bea0f8a667896e10d type value v_measure 69.00410995984436 task dataset metrics type Retrieval type name config split revision C-MTEB/VideoRetrieval MTEB VideoRetrieval default dev 58c2597a5943a2ba48f4668c3b90d796283c5639 type value map_at_1 59.4 type value map_at_10 69.214 type value map_at_100 69.72699999999999 type value map_at_1000 69.743 type value map_at_3 67.717 type value map_at_5 68.782 type value mrr_at_1 59.4 type value mrr_at_10 69.214 type value mrr_at_100 69.72699999999999 type value mrr_at_1000 69.743 type value mrr_at_3 67.717 type value mrr_at_5 68.782 type value ndcg_at_1 59.4 type value ndcg_at_10 73.32300000000001 type value ndcg_at_100 75.591 type value ndcg_at_1000 75.98700000000001 type value ndcg_at_3 70.339 type value ndcg_at_5 72.246 type value precision_at_1 59.4 type value precision_at_10 8.59 type value precision_at_100 0.96 type value precision_at_1000 0.099 type value precision_at_3 25.967000000000002 type value precision_at_5 16.5 type value recall_at_1 59.4 type value recall_at_10 85.9 type value recall_at_100 96.0 type value recall_at_1000 99.1 type value recall_at_3 77.9 type value recall_at_5 82.5 task dataset metrics type Classification type name config split revision C-MTEB/waimai-classification MTEB Waimai default test 339287def212450dcaa9df8c22bf93e9980c7023 type value accuracy 88.53 type value ap 73.56216166534062 type value f1 87.06093694294485

acge model

acge模型来自于合合信息技术团队，对外技术试用平台TextIn, github开源链接为github。合合信息是行业领先的人工智能及大数据科技企业，致力于通过智能文字识别及商业大数据领域的核心技术、C端和B端产品以及行业解决方案为全球企业和个人用户提供创新的数字化、智能化服务。

技术交流请联系yanhui_he@intsig.net，商务合作请联系simon_liu@intsig.net，可以点击图片，扫面二维码来加入我们的微信社群。想加入合合信息，做“文档解析”、“文档检索”、“文档预研”的同学可以投简历给min_du@intsig.net，也可直接添加HR微信详聊岗位内容。

acge是一个通用的文本编码模型，是一个可变长度的向量化模型，使用了Matryoshka Representation Learning，如图所示：

建议使用的维度为1024或者1792

Model Name	Model Size (GB)	Dimension	Sequence Length	Language	Need instruction for retrieval?
acge-text-embedding	0.65	[1024, 1792]	1024	Chinese	NO

Metric

C-MTEB leaderboard (Chinese)

测试的时候因为数据的随机性、显卡、推理的数据类型导致每次推理的结果不一致，我总共测试了4次，不同的显卡(A10 A100)，不同的数据类型，测试结果放在了result文件夹中，选取了一个精度最低的测试作为最终的精度测试。 根据infgrad的建议，选取不用的输入的长度作为测试，Sequence Length为512时测试最佳。

Model Name	GPU	tensor-type	Model Size (GB)	Dimension	Sequence Length	Average (35)	Classification (9)	Clustering (4)	Pair Classification (2)	Reranking (4)	Retrieval (8)	STS (8)
acge_text_embedding	NVIDIA TESLA A10	bfloat16	0.65	1792	1024	68.91	72.76	58.22	87.82	67.67	72.48	62.24
acge_text_embedding	NVIDIA TESLA A100	bfloat16	0.65	1792	1024	68.91	72.77	58.35	87.82	67.53	72.48	62.24
acge_text_embedding	NVIDIA TESLA A100	float16	0.65	1792	1024	68.99	72.76	58.68	87.84	67.89	72.49	62.24
acge_text_embedding	NVIDIA TESLA A100	float32	0.65	1792	1024	68.98	72.76	58.58	87.83	67.91	72.49	62.24
acge_text_embedding	NVIDIA TESLA A100	float16	0.65	1792	768	68.95	72.76	58.68	87.84	67.86	72.48	62.07
acge_text_embedding	NVIDIA TESLA A100	float16	0.65	1792	512	69.07	72.75	58.7	87.84	67.99	72.93	62.09

Reproduce our results

C-MTEB:

import torch
import argparse
import functools
from C_MTEB.tasks import *
from typing import List, Dict
from sentence_transformers import SentenceTransformer
from mteb import MTEB, DRESModel


class RetrievalModel(DRESModel):
    def __init__(self, encoder, **kwargs):
        self.encoder = encoder

    def encode_queries(self, queries: List[str], **kwargs) -> np.ndarray:
        input_texts = ['{}'.format(q) for q in queries]
        return self._do_encode(input_texts)

    def encode_corpus(self, corpus: List[Dict[str, str]], **kwargs) -> np.ndarray:
        input_texts = ['{} {}'.format(doc.get('title', ''), doc['text']).strip() for doc in corpus]
        input_texts = ['{}'.format(t) for t in input_texts]
        return self._do_encode(input_texts)

    @torch.no_grad()
    def _do_encode(self, input_texts: List[str]) -> np.ndarray:
        return self.encoder.encode(
            sentences=input_texts,
            batch_size=512,
            normalize_embeddings=True,
            convert_to_numpy=True
        )


def get_args():
    parser = argparse.ArgumentParser()
    parser.add_argument('--model_name_or_path', default="acge_text_embedding", type=str)
    parser.add_argument('--task_type', default=None, type=str)
    parser.add_argument('--pooling_method', default='cls', type=str)
    parser.add_argument('--output_dir', default='zh_results',
                        type=str, help='output directory')
    parser.add_argument('--max_len', default=1024, type=int, help='max length')
    return parser.parse_args()


if __name__ == '__main__':
    args = get_args()
    encoder = SentenceTransformer(args.model_name_or_path).half()
    encoder.encode = functools.partial(encoder.encode, normalize_embeddings=True)
    encoder.max_seq_length = int(args.max_len)

    task_names = [t.description["name"] for t in MTEB(task_types=args.task_type,
                                                      task_langs=['zh', 'zh-CN']).tasks]
    TASKS_WITH_PROMPTS = ["T2Retrieval", "MMarcoRetrieval", "DuRetrieval", "CovidRetrieval", "CmedqaRetrieval",
                          "EcomRetrieval", "MedicalRetrieval", "VideoRetrieval"]
    for task in task_names:
        evaluation = MTEB(tasks=[task], task_langs=['zh', 'zh-CN'])
        if task in TASKS_WITH_PROMPTS:
            evaluation.run(RetrievalModel(encoder), output_folder=args.output_dir, overwrite_results=False)
        else:
            evaluation.run(encoder, output_folder=args.output_dir, overwrite_results=False)

Usage

acge 中文系列模型

在sentence-transformer库中的使用方法：

from sentence_transformers import SentenceTransformer

sentences = ["数据1", "数据2"]
model = SentenceTransformer('acge_text_embedding')
print(model.max_seq_length)
embeddings_1 = model.encode(sentences, normalize_embeddings=True)
embeddings_2 = model.encode(sentences, normalize_embeddings=True)
similarity = embeddings_1 @ embeddings_2.T
print(similarity)

在sentence-transformer库中的使用方法，选取不同的维度：

from sklearn.preprocessing import normalize
from sentence_transformers import SentenceTransformer

sentences = ["数据1", "数据2"]
model = SentenceTransformer('acge_text_embedding')
embeddings = model.encode(sentences, normalize_embeddings=False)
matryoshka_dim = 1024
embeddings = embeddings[..., :matryoshka_dim]  # Shrink the embedding dimensions
embeddings = normalize(embeddings, norm="l2", axis=1)
print(embeddings.shape)
# => (2, 1024)