Sync model card from docs/model_cards/qwen14b-code-trainer-v9_mixed.md

Comprehensive V9 model card with eval results, training details, and next steps

V10 tool-calling eval (nexus-aligned): 13/14 (93%)

V10 agent eval (nexus-aligned): 7/7 making progress

Phase 4 benchmark: gsm8k (0-shot)

Phase 4 eval-only: phase4-eval-v7.json

Phase 4 benchmark: gsm8k (0-shot)

Phase 4 eval-only: phase4-eval-v7.json

V7 tool-calling eval (nexus-aligned): 8/10 (80%)

V7 agent eval (nexus-aligned): 5/5 making progress

Phase 4 V9 v9_mixed curriculum — eval_loss=0.4550703763961792

initial commit