Hi,

At a glance

The community member is trying to use a custom embedding provided by an embedding server via HTTPS. They found documentation on how to use custom embeddings, but it did not show how to connect to something on the internet. The community members discuss how to send a POST request to the URL and check what is being sent. They provide example code, but the community member is encountering an issue where the code is trying to reach Hugging Face instead of the custom embedding server. The community members are unable to provide a clear answer to the issue.

Useful resources

PPanLyzwa

Hi,
I want to use a custom embeddings provided by embeding servet via https. I found in documentation https://docs.llamaindex.ai/en/stable/examples/embeddings/custom_embeddings/ , but it didnt show how to connect it to something in internet. Does anyone have example?

11 comments

WWhiteFang_Jr

You can do something like this:

Plain Text

    def _get_query_embedding(self, query: str) -> List[float]:
        # Send the POST request, data will be as per your embedding server standard
response = requests.post(url, json=data)
        embeddings = response.json()
        return embeddings

PPanLyzwa

thanks. one more question. how i can check what will be sent to URL?

WWhiteFang_Jr

You can have your own logger and log it what is being sent to the URL

PPanLyzwa

but what is what i will send to URL, is it data?

WWhiteFang_Jr

Yes, but it will totally depend on the URL on how it takes the data

WWhiteFang_Jr

some takes it as JSON body and some take it as param

PPanLyzwa

thanks

PPanLyzwa

I just try it with code:

def _get_query_embedding(self, query: str) -> List[float]:
url = "https:ddddd.com"
headers = {
"Content-Type": "application/json",
"Api-Key": "xxxxxx",
}
data = {
"text": "Ala ma kota"
}

response = requests.post(url, headers=headers, json=data, verify=False)
response.json()
return embeddings[0]

PPanLyzwa

im calling code like:
embed_model = InstructorEmbeddings(embed_batch_size=2)

PPanLyzwa

and as output I have:

No sentence-transformers model found with name sentence-transformers/hub. Creating a new one with mean pooling.

OSError: We couldn't connect to 'https://huggingface.co' to load this file, couldn't find it in the cached files and it looks like sentence-transformers/hub is not the path to a directory containing a file named config.json.
Checkout your internet connection or see how to run the library in offline mode at 'https://huggingface.co/docs/transformers/installation#offline-mode'.

PPanLyzwa

looks like it is ignoring code and trying to reach hugging face, can you help?

Add a reply

Find answers from the community

Hi,