Fix DLRM v2 edge CPU model download to preserve .snapshot_metadata

[Copilot]

🧾 PR Checklist

• Target branch is dev

📌 Note: PRs must be raised against dev. Do not commit directly to main.

📁 File Hygiene & Output Handling

• No unintended files (e.g., logs, cache, temp files, pycache, output folders) are committed

📝 Comments & Communication

• Proper inline comments are added to explain important or non-obvious changes
• PR title and description clearly state what the PR does and why
• Related issues (if any) are properly referenced (Fixes #, Related to #, etc.)

The DLRM_V2 edge CPU path was intermittently failing because the downloaded model_weights payload did not include hidden snapshot files (notably .snapshot_metadata). This updates the r2-downloader weight-sharded variant to use the same packaged download/extract flow that preserves the full model directory contents.

• Change summary
  • get-ml-model-dlrm-terabyte/meta.yaml
    • Updated pytorch,fp32,weight_sharded,r2-downloader env to:
      • enable extraction (MLC_DAE_EXTRACT_DOWNLOADED: 'yes')
      • force zip handling (MLC_EXTRACT_UNZIP: 'yes')
      • use deterministic filename (MLC_DOWNLOAD_FILENAME: download)
      • switch package URL to cloud bundle download endpoint
• Effect
  • model_weights now comes from an extracted archive, preserving hidden metadata files required by the DLRM v2 edge CPU flow.

pytorch,fp32,weight_sharded,r2-downloader:
  env:
    MLC_DAE_EXTRACT_DOWNLOADED: 'yes'
    MLC_DOWNLOAD_FILENAME: download
    MLC_EXTRACT_UNZIP: 'yes'
    MLC_PACKAGE_URL: https://cloud.mlcommons.org/index.php/s/XzfSeLgW8FYfR3S/download

🛡️ Safety & Security

• No secrets or credentials are committed
• Paths, shell commands, and environment handling are safe and portable

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

• inference.mlcommons-storage.org
  • Triggering command: /usr/bin/curl curl -sS -kI --max-redirs 0 --retry 3 --retry-connrefused --connect-timeout 3 --max-time 12 REDACTED (dns block)
  • Triggering command: /usr/bin/curl curl -L --max-time 30 REDACTED (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

• Configure Actions setup steps to set up my environment, which run before the firewall is enabled
• Add the appropriate URLs or hosts to the custom allowlist in this repository's Copilot coding agent settings (admins only)