# Dimensions in hkunlp/instructor-large

**URL:** <https://discuss.huggingface.co/t/dimensions-in-hkunlp-instructor-large/35997>\
**Category:** Models\
**Created:** [April 8, 2023, 11:51pm UTC](https://discuss.huggingface.co/t/dimensions-in-hkunlp-instructor-large/35997 "2023-04-08T23:51:35Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![tiagofreitas87](https://avatars.discourse-cdn.com/v4/letter/t/e36b37/32.png) [@tiagofreitas87](https://discuss.huggingface.co/u/tiagofreitas87)\
**Post date:** [April 8, 2023, 11:51pm UTC](https://discuss.huggingface.co/t/dimensions-in-hkunlp-instructor-large/35997/1 "2023-04-08T23:51:35Z")

</div>

Hi anyone knows how many dimensions does a vector have in instruct-large?

---

<div class="post-metadata">

**Author:** ![pwilczewski](https://avatars.discourse-cdn.com/v4/letter/p/49beb7/32.png) [@pwilczewski](https://discuss.huggingface.co/u/pwilczewski)\
**Post date:** [August 18, 2023, 11:47am UTC](https://discuss.huggingface.co/t/dimensions-in-hkunlp-instructor-large/35997/2 "2023-08-18T11:47:03Z")

</div>

768

I couldn’t find this answer online and Bing hallucinated one. So I just downloaded the model and ran it from langchain. Here are all the details:

HuggingFaceInstructEmbeddings(client=INSTRUCTOR(  
(0): Transformer({‘max\_seq\_length’: 512, ‘do\_lower\_case’: False}) with Transformer model: T5EncoderModel  
(1): Pooling({‘word\_embedding\_dimension’: 768, ‘pooling\_mode\_cls\_token’: False, ‘pooling\_mode\_mean\_tokens’: True, ‘pooling\_mode\_max\_tokens’: False, ‘pooling\_mode\_mean\_sqrt\_len\_tokens’: False, ‘pooling\_mode\_weightedmean\_tokens’: False, ‘pooling\_mode\_lasttoken’: False})  
(2): Dense({‘in\_features’: 1024, ‘out\_features’: 768, ‘bias’: False, ‘activation\_function’: ‘torch.nn.modules.linear.Identity’})  
(3): Normalize()  
), model\_name=‘hkunlp/instructor-large’, cache\_folder=None, model\_kwargs={}, encode\_kwargs={}, embed\_instruction='Test query: ', query\_instruction='Test query: ')