qwenjev / scripts /xet_experiment2.py
tchbcb's picture
v0.4.8: safetensors-loss fix + HEAD verification + v0.4.7 OOM-gate sync
1e3c204 verified
Raw History Blame Contribute Delete
2.65 kB
# Round 2: HEAD-check round-1 results, then large-file (LFS/xet path) test.
# Linear code only.
import os, time, urllib.request
tok = open("/content/qwenjev_work_v3/.hf_token").read().strip()
from huggingface_hub import HfApi
api = HfApi(token=tok)
# ---- step 1: verify round-1 files -------------------------------------
for p in ["t4_assets_v3/_xet_test/default.safetensors",
"t4_assets_v3/_xet_test/noxet.safetensors"]:
url = "https://huggingface.co/tchbcb/qwenjev/resolve/main/" + p
try:
r = urllib.request.urlopen(urllib.request.Request(url, method="HEAD"), timeout=30)
print("R1", p.split("/")[-1], "OK len=" + str(r.headers.get("Content-Length")), flush=True)
except Exception as e:
print("R1", p.split("/")[-1], "FAIL " + str(e)[:40], flush=True)
# ---- step 2: 30MB file => LFS/xet path --------------------------------
big = os.urandom(30 * 1024 * 1024)
open("/tmp/big.safetensors", "wb").write(big)
print("big file:", len(big), "bytes", flush=True)
pA = "t4_assets_v3/_xet_test/big_default.safetensors"
t0 = time.time()
try:
api.upload_file(path_or_fileobj="/tmp/big.safetensors", path_in_repo=pA,
repo_id="tchbcb/qwenjev", repo_type="model",
commit_message="xet test C big default")
print("C upload ok in", round(time.time() - t0, 1), "s", flush=True)
except Exception as e:
print("C upload error:", str(e)[:150], flush=True)
time.sleep(2)
urlA = "https://huggingface.co/tchbcb/qwenjev/resolve/main/" + pA
try:
r = urllib.request.urlopen(urllib.request.Request(urlA, method="HEAD"), timeout=30)
print("C big-default HEAD: OK len=" + str(r.headers.get("Content-Length")), flush=True)
except Exception as e:
print("C big-default HEAD: FAIL " + str(e)[:60], flush=True)
os.environ["HF_HUB_DISABLE_XET"] = "1"
pB = "t4_assets_v3/_xet_test/big_noxet.safetensors"
t0 = time.time()
try:
api.upload_file(path_or_fileobj="/tmp/big.safetensors", path_in_repo=pB,
repo_id="tchbcb/qwenjev", repo_type="model",
commit_message="xet test D big no-xet")
print("D upload ok in", round(time.time() - t0, 1), "s", flush=True)
except Exception as e:
print("D upload error:", str(e)[:150], flush=True)
time.sleep(2)
urlB = "https://huggingface.co/tchbcb/qwenjev/resolve/main/" + pB
try:
r = urllib.request.urlopen(urllib.request.Request(urlB, method="HEAD"), timeout=30)
print("D big-noxet HEAD: OK len=" + str(r.headers.get("Content-Length")), flush=True)
except Exception as e:
print("D big-noxet HEAD: FAIL " + str(e)[:60], flush=True)
print("EXPERIMENT2_DONE", flush=True)