Update main.py

main.py

@@ -12,7 +12,6 @@ app = Flask(__name__)
 n_gpu_layers = 0
 n_batch = 1024
 
-
 llm = LlamaCpp(
     model_path="Phi-3-mini-4k-instruct-q4.gguf",  # path to GGUF file
     temperature=0.1,
@@ -21,23 +20,22 @@ llm = LlamaCpp(
     verbose=True,
     n_ctx=4096
 )
-model0 = AutoModel.from_pretrained('sentence-transformers/paraphrase-TinyBERT-L6-v2', force_download=True)
+model0 = AutoModel.from_pretrained('sentence-transformers/paraphrase-TinyBERT-L6-v2')
 
 model = SentenceTransformer('sentence-transformers/paraphrase-TinyBERT-L6-v2')
 
 file_size = os.stat('Phi-3-mini-4k-instruct-q4.gguf')
 print("model size ====> :", file_size.st_size, "bytes")
 
-
 @app.route('/cv', methods=['POST'])
 def get_skills():
     cv_body = request.json.get('cv_body')
 
     # Simple inference example
     output = llm(
-        f"<|user|>\n{cv_body}<|end|>\n<|assistant|>Can you list the skills mentioned in the CV?<|end|>",
+        f"\n{cv_body}\nCan you list the skills mentioned in the CV?",
         max_tokens=256,  # Generate up to 256 tokens
-        stop=["<|end|>"], 
+        stop=[""], 
         echo=True,  # Whether to echo the prompt
     )
 
@@ -47,13 +45,15 @@ def get_skills():
 def health():
     return jsonify({'status': 'Worked'})
 
-# we will make here post request to compare between lists of skills one has employee just one text and the other has the of jobs has many texts
-# the llm will say the most similar job to the cv
+# Endpoint to compare between employee skills and job skills
 @app.route('/compare', methods=['POST'])
 def compare():
     employee_skills = request.json.get('employee_skills')
     jobs_skills = request.json.get('jobs_skills')
 
+    if not isinstance(employee_skills, list) or not all(isinstance(skill, str) for skill in employee_skills):
+        raise ValueError("employee_skills must be a list of strings")
+
     if not isinstance(jobs_skills, list) or not all(isinstance(skill, str) for skill in jobs_skills):
         raise ValueError("jobs_skills must be a list of strings")
 
@@ -69,18 +69,21 @@ def compare():
         similarity_scores.append({"job": jobs_skills[i], "similarity_score": similarity_score.item()})
 
     return jsonify(similarity_scores)
+
+# Endpoint to compare job posts with employee skills
 @app.route('/compare_jop', methods=['POST'])
 def compare_jop():
     employee_skills = request.json.get('post')
     jobs_skills = request.json.get('employee_skills')
 
     if not isinstance(employee_skills, list) or not all(isinstance(skill, str) for skill in employee_skills):
-        raise ValueError("post must be a list of strings")
-    if not isinstance(jobs_skills, list) or not all(isinstance(skill, str) for skill in jobs_skills):
         raise ValueError("employee_skills must be a list of strings")
 
-    employee_embeddings = model.encode(employee_skills)
+    if not isinstance(jobs_skills, list) or not all(isinstance(skill, str) for skill in jobs_skills):
+        raise ValueError("jobs_skills must be a list of strings")
+
     job_embeddings = model.encode(jobs_skills)
+    employee_embeddings = model.encode(employee_skills)
 
     similarity_scores = []
     employee_embeddings_tensor = torch.from_numpy(employee_embeddings).unsqueeze(0)
@@ -92,7 +95,5 @@ def compare_jop():
 
     return jsonify(similarity_scores)
 
-
-
 if __name__ == '__main__':
-    app.run()
+    app.run()