Update app.py

app.py

@@ -25,10 +25,9 @@ from io import StringIO
 last = 0
 CHROMA_DATA_PATH = "chroma_data/"
 EMBED_MODEL =  "sentence-transformers/all-MiniLM-L6-v2" # "BAAI/bge-m3"
-LLM_NAME = "mistralai/Mistral-Nemo-Instruct-2407"
-LLM_NAME = "swiss-ai/Apertus-8B-Instruct-2509"
+#LLM_NAME = "mistralai/Mistral-Nemo-Instruct-2407"
+LLM_NAME = "swiss-ai/Apertus-8B-Instruct-2509:publicai"
 #LLM_NAME = "W4D/YugoGPT-7B-Instruct-GGUF"
-# all-MiniLM-L6-v2
 CHUNK_SIZE = 800
 CHUNK_OVERLAP = 50
 max_results = 3
@@ -77,9 +76,10 @@ Settings.llm = HuggingFaceInferenceAPI(model_name=LLM_NAME,
 #                              stopping_ids=[50278, 50279, 50277, 1, 0],
                               generate_kwargs={"temperature": temperature, "top_p":top_p, "repetition_penalty": repetition_penalty,
                                                "presence_penalty": presence_penalty, "frequency_penalty": frequency_penalty,
-                                               "top_k": top_k, "do_sample": False, "task": None},
+                                               "top_k": top_k, "do_sample": False },
 #                              tokenizer_kwargs={"max_length": 4096},
                               tokenizer_name=LLM_NAME,
+                              api_key="hf_" + HF_TOKEN,
 #                              provider="auto",
 #                              task="None"
                              )