Spaces:
Sleeping
Sleeping
File size: 2,178 Bytes
2534516 db7aed9 f62e4fd db7aed9 27d2b1f dc11f1f db7aed9 f62e4fd dc11f1f 27d2b1f dc11f1f db7aed9 2534516 2a28615 db7aed9 27d2b1f db7aed9 27d2b1f dc11f1f 2a28615 db7aed9 27d2b1f 2534516 db7aed9 f62e4fd |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 |
import torch
from flask import Flask, request, jsonify
from langchain_community.llms import LlamaCpp
from sentence_transformers import SentenceTransformer
from transformers import AutoModel
import os
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,
n_gpu_layers=n_gpu_layers,
n_batch=n_batch,
verbose=True,
n_ctx=4096
)
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")
# Assume 'model' is already defined somewhere else
@app.route('/compare', methods=['POST'])
def compare():
data = request.json
# Validation and data extraction
if 'jobs_skills' not in data or 'employee_skills' not in data:
return jsonify({"error": "Missing 'jobs_skills' or 'employee_skills' in request"}), 400
jobs_skills = data['jobs_skills']
employee_skills = data['employee_skills']
if not isinstance(jobs_skills, list) or not all(isinstance(skill, str) for skill in jobs_skills):
return jsonify({"error": "'jobs_skills' must be a list of strings"}), 400
if not isinstance(employee_skills, list) or not all(isinstance(skill, str) for skill in employee_skills):
return jsonify({"error": "'employee_skills' must be a list of strings"}), 400
# Encoding skills into embeddings
job_embeddings = model.encode(jobs_skills)
employee_embeddings = model.encode(employee_skills)
# Computing cosine similarity between employee skills and each job
similarity_scores = []
employee_embeddings_tensor = torch.from_numpy(employee_embeddings).unsqueeze(0)
for i, job_e in enumerate(job_embeddings):
job_e_tensor = torch.from_numpy(job_e).unsqueeze(0)
similarity_score = cosine_similarity(employee_embeddings_tensor, job_e_tensor, dense_output=True)
similarity_scores.append({"job": jobs_skills[i], "similarity_score": similarity_score.item()})
return jsonify(similarity_scores)
if __name__ == '__main__':
app.run(debug=True)
|