Saving and loading bigcode/santacoder model on a mac results in "ModuleNotFoundError: No module named 'transformers_modules'
#29
by kanandk - opened
Code:
from transformers import AutoModelForCausalLM, AutoTokenizer, AutoModel
import torch
model = torch.load("./tmp/gptj.pt")
model.eval()
tokenizer = AutoTokenizer.from_pretrained("./tmp")
device = "cpu"
inputs = tokenizer.encode("function helloWorld():", return_tensors="pt").to(device)
outputs = model.generate(inputs)
print(tokenizer.decode(outputs[0]))
Transformers version : 4.28.1
Torch version : 2.0.0
Python : 3.9.15
Mac M1 with OSX
Could you please share the stack trace.