File size: 512 Bytes
54d04d4 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 | # cache_download.py - Run ONCE
from transformers import AutoModelForCausalLM, AutoTokenizer
import os
MODEL_NAME = "microsoft/Phi-3-mini-4k-instruct"
MODEL_NAME = "Qwen/Qwen2-1.5B-Instruct" # ✅ No tokenizer bugs
print("Caching Phi-3...")
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, cache_dir="./hf_cache")
model = AutoModelForCausalLM.from_pretrained(
MODEL_NAME,
cache_dir="./hf_cache",
torch_dtype="auto",
device_map="cpu"
)
print("✅ Cached to ./hf_cache/")
|