33 lines
1.1 KiB
Python
33 lines
1.1 KiB
Python
#!/usr/bin/env python3
|
|
"""Dump checkpoint key names for specific layers only."""
|
|
import json
|
|
from pathlib import Path
|
|
from safetensors.torch import load_file
|
|
|
|
CHECKPOINT_DIR = "/root/nvidia-meeting/DeepSeek-V4-Pro"
|
|
|
|
def main():
|
|
cdir = Path(CHECKPOINT_DIR)
|
|
index_path = cdir / "model.safetensors.index.json"
|
|
with open(index_path) as f:
|
|
weight_map = json.load(f).get("weight_map", {})
|
|
|
|
# Only show layer 0, 2, and 60 keys (non-expert)
|
|
for li in [0, 2, 60]:
|
|
prefix = f"model.layers.{li}."
|
|
keys = sorted(k for k in weight_map if k.startswith(prefix))
|
|
# Filter out individual expert weights
|
|
filtered = [k for k in keys if '.experts.' not in k]
|
|
print(f"\n=== Layer {li} keys ({len(filtered)} non-expert) ===")
|
|
for k in filtered:
|
|
print(f" {k}")
|
|
|
|
# Non-layer keys (short list)
|
|
other_keys = sorted(k for k in weight_map if not k.startswith("model.layers."))
|
|
print(f"\n=== Non-layer keys ({len(other_keys)}) ===")
|
|
for k in other_keys:
|
|
print(f" {k}")
|
|
|
|
if __name__ == "__main__":
|
|
main()
|