Upload app.py

app.py

@@ -7,17 +7,16 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 
 # Load the gated model
 #model_name = "RickyDeSkywalker/TheoremLlama"
-#model_name = "unsloth/Llama-3.2-1B-Instruct"
+model_name = "unsloth/Llama-3.2-1B-Instruct"
 device = "cuda" if torch.cuda.is_available() else "cpu"
 #model_name = "internlm/internlm2-math-plus-7b"
-model_name = "deepseek-ai/DeepSeek-Prover-V1.5-RL"
+#model_name = "deepseek-ai/DeepSeek-Prover-V1.5-RL"
 HF_TOKEN = os.environ.get("HF_TOKEN")
 #login(HF_TOKEN)
 
 tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
 # Set `torch_dtype=torch.float16` to load model in float16, otherwise it will be loaded as float32 and might cause OOM Error.
 model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=True, torch_dtype=torch.float16).eval().to(device)
-model = model.eval()
 terminators = [tokenizer.eos_token_id, 
                tokenizer.convert_tokens_to_ids("<|eot_id|>"), 
                tokenizer.convert_tokens_to_ids("<|reserved_special_token_26|>")]