Open lucafrost opened 1 year ago
Could you give a sample of what is in the embeddings[1-3]? That may help us work out what is going on.
hey @dtaivpp — appreciate the quick response!
embeddings[0-2]
are 768-dimensional embeddings generated for semantic search to be used with the kNN plugin.
updated example below...
emb1 = [0, 1, 2, ... 768]
emb2 = [0, 1, 2, ... 768]
action = { "update": { "_index": "my_index", "_id": doc_id }}
source = {
"doc": {
"title.vector": emb1,
"description.vector": emb2
}
}
payload = ""
payload += json.dumps(action) + "\n"
payload += json.dumps(source) + "\n"
client.bulk(body=payload)
let me know if that helps at all? 🙇🏼♂️
Without looking at the problem, if anyone has time, it would be helpful to try to get to a smaller/simpler end-to-end repro and turn it into a failing unit test.
Describe the bug When attempting to use the bulk API to update documents, the update is rejected with an illegal argument exception stating
DocValuesField "description.vector" appears more than once in this document (only one value is allowed per field
.To Reproduce See the below Python script...
Plugins All of 'em -- see AWS OSS Plugins, full output from
_cat/plugins
is here.Host/Environment (please complete the following information):