Xenova HF staff commited on
Commit
f2ec3aa
1 Parent(s): a635bdb

Add transformers.js sample code

Browse files
Files changed (1) hide show
  1. README.md +20 -0
README.md CHANGED
@@ -185,6 +185,26 @@ embeddings = model.encode([
185
  print(cos_sim(embeddings[0], embeddings[1]))
186
  ```
187
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
188
  ## Plans
189
 
190
  1. Bilingual embedding models supporting more European & Asian languages, including Spanish, French, Italian and Japanese.
 
185
  print(cos_sim(embeddings[0], embeddings[1]))
186
  ```
187
 
188
+ You can also use the [Transformers.js](https://huggingface.co/docs/transformers.js) library to compute embeddings in JavaScript.
189
+ ```js
190
+ // npm i @xenova/transformers
191
+ import { pipeline, cos_sim } from '@xenova/transformers';
192
+
193
+ const extractor = await pipeline('feature-extraction', 'jinaai/jina-embeddings-v2-base-code', {
194
+ quantized: false, // Comment out this line to use the 8-bit quantized version
195
+ });
196
+
197
+ const texts = [
198
+ 'How do I access the index while iterating over a sequence with a for loop?',
199
+ '# Use the built-in enumerator\nfor idx, x in enumerate(xs):\n print(idx, x)',
200
+ ]
201
+ const embeddings = await extractor(texts, { pooling: 'mean' });
202
+
203
+ const score = cos_sim(embeddings[0].data, embeddings[1].data);
204
+ console.log(score);
205
+ // 0.7281748759529421
206
+ ```
207
+
208
  ## Plans
209
 
210
  1. Bilingual embedding models supporting more European & Asian languages, including Spanish, French, Italian and Japanese.