from datasets import load_dataset
from huggingface_hub import hf_hub_download
import os
D="data"
for name, path, cfg in [
    ("smol_smoltalk","HuggingFaceTB/smol-smoltalk",None),
    ("everyday","HuggingFaceTB/smoltalk","everyday-conversations"),
    ("apigen","argilla/apigen-function-calling",None),
    ("hermes_fc","NousResearch/hermes-function-calling-v1","func_calling"),
    ("hermes_fc_single","NousResearch/hermes-function-calling-v1","func_calling_singleturn"),
    ("hermes_glaive","NousResearch/hermes-function-calling-v1","glaive_func_calling"),
]:
    out=os.path.join(D,name)
    if os.path.exists(out): print("skip",name); continue
    ds=load_dataset(path,cfg) if cfg else load_dataset(path)
    ds.save_to_disk(out); print("saved",name,ds)
hf_hub_download("nvidia/Nemotron-Agentic-v1","data/tool_calling.jsonl",repo_type="dataset",local_dir=os.path.join(D,"nemotron"))