Spaces:
Sleeping
Sleeping
File size: 2,542 Bytes
5d94b57 9f8186b b1711fb 9f8186b 7716f22 9f8186b c9f8cf6 ae4a9b6 9f8186b b1711fb 9f8186b b1711fb 9f8186b b1711fb 9f8186b b1711fb cc17c3b b1711fb cc17c3b b1711fb cc17c3b b1711fb cc17c3b b1711fb cc17c3b b1711fb cc17c3b 549f944 |
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 65 66 67 68 69 70 71 72 73 74 |
from flask import Flask, request, jsonify
from langchain_community.llms import LlamaCpp
from sentence_transformers import SentenceTransformer
from transformers import AutoTokenizer, AutoModel
# cosine_similarity
import torch
from torch.nn.functional import cosine_similarity
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
)
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|>",
max_tokens=256, # Generate up to 256 tokens
stop=["<|end|>"],
echo=True, # Whether to echo the prompt
)
return jsonify({'skills': output})
@app.get('/')
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
@app.route('/compare', methods=['POST'])
def compare():
employee_skills = request.json.get('employee_skills') # CV text
jobs_skills = request.json.get('jobs_skills') # List of job skills
if not isinstance(jobs_skills, list) or not all(isinstance(skill, str) for skill in jobs_skills):
raise ValueError("The jobs_skills must be a list of strings")
# Convert texts to embeddings arrays
employee_embedding = np.array([model.encode(employee_skills)])
job_embeddings = np.array([model.encode(skill) for skill in jobs_skills])
# Calculate similarity using cosine similarity
similarities = cosine_similarity(employee_embedding, job_embeddings)[0]
# Find the most similar job and its corresponding similarity score
max_similarity = np.max(similarities)
most_similar_index = np.argmax(similarities)
most_similar_job = jobs_skills[most_similar_index]
return jsonify({'job': most_similar_job, 'similarity_score': max_similarity})
if __name__ == '__main__':
app.run() |