Update: more logging for model performance

This commit is contained in:
dx-tan 2023-12-29 14:24:52 +07:00
parent 0fafb8f565
commit ad249a8c19
4 changed files with 9 additions and 10 deletions

View File

@ -20,12 +20,12 @@ def aggregate_result(results):
doc_types = []
for index, result in results.items():
index = int(index)
doc_type = result["metadata"]["doc_type"]
doc_type = result.get("metadata", {}).get("doc_type", "all")
doc_types.append(doc_type)
sorted_results[index] = ((doc_type, result))
des_result = deepcopy(list(results.values()))[0]
des_result.pop("metadata")
des_result.pop("metadata", None)
des_result["content"]["total_pages"] = 0
des_result["content"]["ocr_num_pages"] = 0
des_result["content"]["document"][0]["end_page"] = 0
@ -162,15 +162,14 @@ def process_invoice_sbt_result(rq_id, result, metadata):
ai_inference_profile = {}
doc_type_string = ""
for idx, result in results.items():
ai_inference_profile["{doc_type}_{idx}".format(doc_type=result["metadata"]["doc_type"], idx=result["metadata"]["index_in_request"])] = result["metadata"]["ai_inference_profile"]
doc_type_string += "{},".format(result["metadata"]["doc_type"])
ai_inference_profile["{doc_type}_{idx}".format(doc_type=result.get("metadata", {}).get("doc_type", "all"), idx=result.get("metadata", {}).get("index_in_request", 0))] = result.get("metadata", {}).get("ai_inference_profile", {})
doc_type_string += "{},".format(result.get("metadata", {}).get("doc_type", "all"))
doc_type_string = doc_type_string[:-1]
rq.ai_inference_profile = ai_inference_profile
rq.doc_type = doc_type_string
rq.ai_inference_start_time = result["metadata"]["ai_inference_start_time"] # advancing the last result
rq.preprocessing_time = result["metadata"]["preprocessing_time"] # advancing the last result
rq.ai_inference_start_time = result.get("metadata", {}).get("ai_inference_start_time", -1) # advancing the last result
rq.preprocessing_time = result.get("metadata", {}).get("preprocessing_time", 0) # advancing the last result
rq.ai_inference_time = time.time() - rq.ai_inference_start_time
rq.save()
else:

View File

@ -307,7 +307,7 @@ def token_value(token_type):
return 1 # Basic OCR
def send_to_queue2(rq_id, sub_id, file_url, user_id, typez, metadata):
def send_to_queue2(rq_id, sub_id, file_url, user_id, typez, metadata={}):
try:
if typez == ProcessType.ID_CARD.value:
c_connector.process_id(

View File

@ -79,7 +79,7 @@ def deploy():
# Define the variable
tag = str(random_hash()[:8])
now = datetime.now()
tag = tag + "_" + str(now.strftime("%d%m%y%H%M%S"))
tag = str(now.strftime("%Y%m%d%H%M%S")) + "_" + tag
# tag = "4cae5134_261223123256"
print(tag)