yaoqi commited on
Commit
1d1cf06
1 Parent(s): 9213972

Update image-similarity-cn.md

Browse files
Files changed (1) hide show
  1. image-similarity-cn.md +4 -4
image-similarity-cn.md CHANGED
@@ -231,7 +231,7 @@ Top 5 candidate labels: [0, 0, 0, 0, 0]
231
 
232
  如果我们能以某种方式降低嵌入的维度而不影响它们的意义,我们仍然可以在速度和检索质量之间保持良好的折衷。本文 [附带的 Colab Notebook](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/image_similarity.ipynb) 实现并演示了如何通过随机投影 (random projection) 和位置敏感哈希(locality-sensitive hashing,LSH)这两种方法来取得折衷。
233
 
234
- 🤗 Datasets 提供与 [FAISS](https://github.com/facebookresearch/faiss) 的直接集成,进一步简化了构建相似性系统的过程。假设你已经提取了候选图像的嵌入(`beans` 数据集)并把他们存储在称为 `embedding` 的 `feature` 中。你现在可以轻松地使用 `daatset` 的 [`add_faiss_index ()`](https://huggingface.co/docs/datasets/v2.7.1/en/package_reference/main_classes#datasets.Dataset.add_faiss_index) 方法来构建稠密索引:
235
 
236
  ```py
237
  dataset_with_embeddings.add_faiss_index (column="embeddings")
@@ -261,6 +261,6 @@ scores, retrieved_examples = dataset_with_embeddings.get_nearest_examples (
261
  * [ScaNN:高效向量相似性搜索](http://ai.googleblog.com/2020/07/announcing-scann-efficient-vector.html)
262
  * [在移动应用程序中集成图像搜索引擎](https://www.tensorflow.org/lite/inference_with_metadata/task_library/image_searcher)
263
 
264
- >>>> 英文原文: <url> https://huggingface.co/blog/dialog-agents </url>
265
- >>>>
266
- >>>> 译者: Matrix Yao (姚伟峰),英特尔深度学习工程师,工作方向为 transformer-family 模型在各模态数据上的应用及大规模模型的训练推理。
 
231
 
232
  如果我们能以某种方式降低嵌入的维度而不影响它们的意义,我们仍然可以在速度和检索质量之间保持良好的折衷。本文 [附带的 Colab Notebook](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/image_similarity.ipynb) 实现并演示了如何通过随机投影 (random projection) 和位置敏感哈希(locality-sensitive hashing,LSH)这两种方法来取得折衷。
233
 
234
+ 🤗 Datasets 提供与 [FAISS](https://github.com/facebookresearch/faiss) 的直接集成,进一步简化了构建相似性系统的过程。假设你已经提取了候选图像的嵌入(`beans` 数据集)并把他们存储在称为 `embedding` 的 `feature` 中。你现在可以轻松地使用 `dataset` 的 [`add_faiss_index ()`](https://huggingface.co/docs/datasets/v2.7.1/en/package_reference/main_classes#datasets.Dataset.add_faiss_index) 方法来构建稠密索引:
235
 
236
  ```py
237
  dataset_with_embeddings.add_faiss_index (column="embeddings")
 
261
  * [ScaNN:高效向量相似性搜索](http://ai.googleblog.com/2020/07/announcing-scann-efficient-vector.html)
262
  * [在移动应用程序中集成图像搜索引擎](https://www.tensorflow.org/lite/inference_with_metadata/task_library/image_searcher)
263
 
264
+ > 英文原文: <url> https://huggingface.co/blog/dialog-agents </url>
265
+ >
266
+ > 译者: Matrix Yao (姚伟峰),英特尔深度学习工程师,工作方向为 transformer-family 模型在各模态数据上的应用及大规模模型的训练推理。