Upload 2 files

app.py

@@ -0,0 +1,90 @@
+import gradio as gr
+from transformers import pipeline, AutoTokenizer, AutoModel
+import torch
+from sklearn.metrics.pairwise import cosine_similarity
+import numpy as np
+
+device = 0 if torch.cuda.is_available() else -1
+
+# Загружаем модели
+ner_pipeline = pipeline("ner", model="dslim/bert-base-NER", device=device)
+text_generator = pipeline("text-generation", model="gpt2", device=device)
+
+# Sentence Transformer для similarity
+tokenizer = AutoTokenizer.from_pretrained("sentence-transformers/all-MiniLM-L6-v2")
+model = AutoModel.from_pretrained("sentence-transformers/all-MiniLM-L6-v2")
+if torch.cuda.is_available():
+    model = model.cuda()
+
+def get_embedding(text):
+    inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=512, padding=True)
+    if torch.cuda.is_available():
+        inputs = {k: v.cuda() for k, v in inputs.items()}
+    with torch.no_grad():
+        outputs = model(**inputs)
+    return outputs.last_hidden_state.mean(dim=1).cpu().numpy()
+
+def calculate_similarity(resume, job):
+    resume_emb = get_embedding(resume)
+    job_emb = get_embedding(job)
+    similarity = cosine_similarity(resume_emb, job_emb)[0][0]
+    return similarity
+
+def extract_skills(text):
+    entities = ner_pipeline(text)
+    skills = []
+    for ent in entities:
+        if ent['entity'].endswith('MISC') or ent['entity'].endswith('ORG'):
+            skills.append(ent['word'])
+    return list(set(skills))
+
+def generate_recommendations(resume_skills, job_skills):
+    missing = [s for s in job_skills if s not in resume_skills]
+    if not missing:
+        return "Все необходимые навыки присутствуют"
+    
+    prompt = f"Add these skills to resume: {', '.join(missing[:3])}"
+    result = text_generator(prompt, max_length=50, num_return_sequences=1, pad_token_id=50256)
+    return result[0]['generated_text']
+
+def analyze(resume_text, job_text):
+    if not resume_text or not job_text:
+        return "Заполните оба поля", "", ""
+    
+    # Similarity
+    similarity = calculate_similarity(resume_text, job_text)
+    match_score = f"Match Score: {similarity*100:.1f}%"
+    
+    # Skills extraction
+    resume_skills = extract_skills(resume_text)
+    job_skills = extract_skills(job_text)
+    missing_skills = [s for s in job_skills if s not in resume_skills]
+    skills_gap = f"Missing: {', '.join(missing_skills[:5]) if missing_skills else 'None'}"
+    
+    # Recommendations
+    recommendations = generate_recommendations(resume_skills, job_skills)
+    
+    return match_score, skills_gap, recommendations
+
+with gr.Blocks() as demo:
+    gr.Markdown("# Resume-Job Matcher")
+    gr.Markdown("Система сопоставления резюме и вакансий")
+    
+    with gr.Row():
+        resume_input = gr.Textbox(label="Resume", lines=10, placeholder="Paste resume here...")
+        job_input = gr.Textbox(label="Job Description", lines=10, placeholder="Paste job description here...")
+    
+    analyze_btn = gr.Button("Analyze")
+    
+    with gr.Row():
+        match_output = gr.Textbox(label="Match Score")
+        skills_output = gr.Textbox(label="Skills Gap")
+        recommendations_output = gr.Textbox(label="Recommendations")
+    
+    analyze_btn.click(
+        fn=analyze,
+        inputs=[resume_input, job_input],
+        outputs=[match_output, skills_output, recommendations_output]
+    )
+
+demo.launch()

requirements.txt

@@ -0,0 +1,6 @@
+transformers
+torch
+gradio
+scikit-learn
+numpy
+sentence-transformers