Spaces:
Sleeping
Sleeping
File size: 1,133 Bytes
d9ded01 b2c1928 ea0af80 b2c1928 d9ded01 b2c1928 8374669 d36dc81 ba41c7f d9ded01 ba41c7f ad9f174 ba41c7f d36dc81 ad9f174 eb66cb5 ad9f174 ea0af80 ad9f174 b2c1928 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 |
import os
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
MODEL_NAME = "bigcode/starcoderbase-1b"
HF_TOKEN = os.getenv("HUGGINGFACE_TOKEN")
# Force CPU mode
device = "cpu"
# Load tokenizer and model
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, token=HF_TOKEN)
# Ensure the tokenizer has a pad token set
if tokenizer.pad_token is None:
tokenizer.pad_token = tokenizer.eos_token # Set pad_token to eos_token
model = AutoModelForCausalLM.from_pretrained(
MODEL_NAME,
token=HF_TOKEN,
torch_dtype=torch.float32, # Use float32 for CPU
trust_remote_code=True
).to(device) # Move model explicitly to CPU
def generate_code(prompt: str, max_tokens: int = 256):
inputs = tokenizer(
prompt,
return_tensors="pt",
padding=True,
truncation=True, # Allow truncation
max_length=1024 # Set a maximum length explicitly
).to(device)
output = model.generate(
**inputs,
max_new_tokens=max_tokens,
pad_token_id=tokenizer.pad_token_id
)
return tokenizer.decode(output[0], skip_special_tokens=True)
|