Hugging Face Hub, XET debugging

Debugging, troubleshooting stalled, stuck downloads and slow downloads

Downloads are stuck at 90% to 99%

If you see downloads via hf download unsloth/* get stuck at 90% or 99% of progress for quite some time, cancel the current run, and try adding using below commands:

pip install -U huggingface_hub

HF_HOME=".cache_new/huggingface" \
HF_XET_CACHE=".cache_new/huggingface/xet" \
HF_HUB_CACHE=".cache_new/huggingface/hub" \
HF_XET_HIGH_PERFORMANCE=1 \
HF_XET_CHUNK_CACHE_SIZE_BYTES=0 \
HF_XET_RECONSTRUCT_WRITE_SEQUENTIALLY=0 \
HF_XET_NUM_CONCURRENT_RANGE_GETS=64 \
hf download unsloth/Qwen3-Coder-Next-GGUF \
    --local-dir unsloth/Qwen3-Coder-Next-GGUF \
    --include "*UD-Q6_K_XL*"

Rate limited or 429 Too Many Requests?

Try using snapshot_download instead, then import Unsloth which will set the correct Hugging Face variables for you:

import unsloth
import os

os.environ["HF_HOME"] = ".cache_new/huggingface"
os.environ["HF_XET_CACHE"] = ".cache_new/huggingface/xet"
os.environ["HF_HUB_CACHE"] = ".cache_new/huggingface/hub"
from huggingface_hub import snapshot_download
snapshot_download(
    repo_id = "unsloth/Qwen3-Coder-Next-GGUF",
    local_dir = "unsloth/Qwen3-Coder-Next-GGUF",
    allow_patterns = ["*UD-Q6_K_XL*"],
)

Or maybe try getting a Hugging Face token first via https://huggingface.co/settings/tokens

pip install -U huggingface_hub

HF_HOME=".cache_new/huggingface" \
HF_XET_CACHE=".cache_new/huggingface/xet" \
HF_HUB_CACHE=".cache_new/huggingface/hub" \
HF_XET_HIGH_PERFORMANCE=1 \
HF_XET_CHUNK_CACHE_SIZE_BYTES=0 \
HF_XET_RECONSTRUCT_WRITE_SEQUENTIALLY=0 \
HF_XET_NUM_CONCURRENT_RANGE_GETS=64 \
    hf download unsloth/Qwen3-Coder-Next-GGUF \
    --local-dir unsloth/Qwen3-Coder-Next-GGUF \
    --include "*UD-Q6_K_XL*" \
    --token "hf_ADD_YOUR_HUGGING_FACE_TOKEN_HERE"