reasoning-gym/training/utils/load_fsdp_to_hf.py
2025-07-28 15:57:15 +01:00

58 lines
2 KiB
Python

#!/usr/bin/env python
# encoding: utf-8
from collections import defaultdict
from glob import glob
import fire
import torch
from huggingface_hub import HfApi, create_repo
from transformers import AutoConfig, AutoModelForCausalLM, AutoTokenizer
def main(fsdp_checkpoint_path, huggingface_model_path, output_path, push_to_hub=True, hub_token="", private=True):
state_dict = defaultdict(list)
world_size = 4
for rank in range(world_size):
filepath = f"{fsdp_checkpoint_path}/model_world_size_{world_size}_rank_{rank}.pt"
print("loading", filepath)
this_state_dict = torch.load(filepath)
for key, value in this_state_dict.items():
state_dict[key].append(value.to_local())
for key in state_dict:
state_dict[key] = torch.cat(state_dict[key], dim=0)
config = AutoConfig.from_pretrained(huggingface_model_path)
model = AutoModelForCausalLM.from_config(config)
model.load_state_dict(state_dict)
model.save_pretrained(output_path, max_shard_size="10GB")
tokenizer = AutoTokenizer.from_pretrained(huggingface_model_path)
tokenizer.save_pretrained(output_path)
# Push to hub if requested
if push_to_hub:
if not output_path:
raise ValueError("output path must be provided when push_to_hub=True")
print(f"Pushing model to Hugging Face Hub: {output_path}")
# Create repository if it doesn't exist
api = HfApi(token=hub_token)
try:
create_repo(repo_id=output_path, private=private, exist_ok=True, token=hub_token)
print(f"Repository {output_path} created or already exists")
except Exception as e:
print(f"Repository creation info: {e}")
# Push model and tokenizer to hub
model.push_to_hub(output_path, token=hub_token, private=private)
tokenizer.push_to_hub(output_path, token=hub_token, private=private)
print(f"✅ Model successfully pushed to https://huggingface.co/{output_path}")
if __name__ == "__main__":
fire.Fire(main)