Add validation for LabelAttentionClassifier head configuration

[Copilot]

LabelAttentionClassifier was missing validation for head configuration constraints, allowing invalid configurations that cause shape mismatches at runtime when embedding_dim is not divisible by n_head.

Changes

• Added early validation in LabelAttentionClassifier.__init__:
  • embedding_dim must be divisible by n_head (prevents shape mismatch in c_proj layer)
  • n_head must be divisible by n_kv_head (required for Group Query Attention)
• Follows existing TextEmbedder validation pattern for consistency
• Error messages include integer values and remainders for clarity

Example

Invalid configurations now fail fast with clear error messages:

config = TextEmbedderConfig(
    embedding_dim=127,  # Not divisible by 8
    label_attention_config=LabelAttentionConfig(n_head=8, n_kv_head=4, num_classes=5)
)
# Raises: ValueError: embedding_dim (127) must be divisible by n_head (8). 
#         Got head_dim = 15 with remainder 7

Previously this would fail later with cryptic tensor shape errors in c_proj.

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[Copilot]

Just as a heads up, I was blocked by some firewall rules while working on your feedback. Expand below for details.

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:

• astral.sh
  • Triggering command: /usr/bin/curl curl -LsSf 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)