Store Message embedding #540
Conversation
| @@ -41,6 +51,10 @@ async def post(self, input: str) -> Any: | |||
| async with session.post(self.api_url, headers=self.headers, json=payload) as response: | |||
| # If we get a bad response | |||
| if response.status != 200: | |||
| from loguru import logger | |||
| @@ -95,6 +95,7 @@ services: | |||
| - DEBUG_SKIP_API_KEY_CHECK=True | |||
| - DEBUG_USE_SEED_DATA=True | |||
| - MAX_WORKERS=1 | |||
| - DEBUG_SKIP_EMBEDDING_COMPUTATION=True | |||
There was a problem hiding this comment.
could you also add this to the ansible playbook?
There was a problem hiding this comment.
When deployed via ansible the fetching of the embeddings should not be skipped?
There was a problem hiding this comment.
When should be skipped? Only in local-run.sh?
There was a problem hiding this comment.
That's why I asked again... probably also won't be necessary for Frontend development..
| @@ -41,6 +51,10 @@ async def post(self, input: str) -> Any: | |||
| async with session.post(self.api_url, headers=self.headers, json=payload) as response: | |||
There was a problem hiding this comment.
do we know what happens if the text is longer than the intended text length of the model?
There was a problem hiding this comment.
I am not sure about this.
If there is a certain sentence length pre-defined in the model, the extra words will be truncated.
And the arch of our model is the following:
SentenceTransformer(
(0): Transformer({
'max_seq_length': 128,
'do_lower_case': False
}) with Transformer model: BertModel
(1): Pooling({
'word_embedding_dimension': 384,
'pooling_mode_cls_token': False,
'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False,
'pooling_mode_mean_sqrt_len_tokens': False})
)So would have a sentence length of 128 words. The impact on us will depend then if there is an important amount of messages that have > 128 words.
Related Issue here.
nil-andreu
force-pushed
the
messageEmbeddings
branch
from
877c744 to
ef7bd89
Compare
Yes, all of the text processing is done by the endpoint. And we are also storing which is the version that we are using of that model.
Okay, have applied also those changes. |
Thankss! |
Create message embeddings and store them in the DB.
Related to this issue.