ss

app.py

@@ -19,6 +19,8 @@ import PyPDF2
 from docx import Document
 import csv
 from explanation_generator import ExplanationGenerator
+from einops.layers.torch import Rearrange, Reduce
+from einops import rearrange, reduce, repeat
 
 # Download NLTK resources
 try:
@@ -366,8 +368,11 @@ class ResumeScreener:
         bm25_scores = self.calculate_bm25_scores(resume_texts, job_description)
         
         # Normalize BM25 scores
-        if max(bm25_scores) > 0:
+        if bm25_scores and len(bm25_scores) > 0 and max(bm25_scores) > 0:
             bm25_scores = [score / max(bm25_scores) for score in bm25_scores]
+        else:
+            # If BM25 scores are empty or all zero, use neutral values
+            bm25_scores = [0.5] * len(resume_texts)
         
         # Calculate hybrid scores
         keyword_weight = 1.0 - semantic_weight

explanation_generator.py

@@ -9,6 +9,8 @@ import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
 import os
 import re
+from einops.layers.torch import Rearrange, Reduce
+from einops import rearrange, reduce, repeat
 
 # Load QwQ model at initialization time
 print("Loading Qwen/QwQ-32B model with 4-bit quantization...")

requirements.txt

@@ -18,3 +18,4 @@ huggingface-hub>=0.30.0,<1.0
 einops
 bitsandbytes>=0.41.0
 accelerate>=0.23.0
+replicate==0.17.0