ok

app.py

@@ -7,14 +7,13 @@ from docx import Document
 from docx.shared import Pt
 from docx.oxml.ns import qn
 
-# Thiết lập token API từ Hugging Face
-import os
-os.environ["HUGGINGFACEHUB_API_TOKEN"] = "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"
+# Thiết lập API token từ Hugging Face
+api_token = "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"
 
 # Tải mô hình và tokenizer từ Hugging Face
 model_name = "meta-llama/Meta-Llama-3.1-8B-Instruct"
-tokenizer = AutoTokenizer.from_pretrained(model_name, use_auth_token=True)
-model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16, use_auth_token=True).to("cuda")
+tokenizer = AutoTokenizer.from_pretrained(model_name, token=api_token)
+model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16, token=api_token).to("cuda")
 
 def analyze_chat(link, extraction_request, system_message, max_tokens, temperature, top_p):
     response = requests.get(link)

requirements.txt

@@ -2,4 +2,5 @@ gradio
 transformers
 requests
 pdfkit
-python-docx
+python-docx
+numpy==1.21.0