router-expert-code.algorithms
LoRA expert for the code.algorithms target of the route-then-admit pool, trained on meirdick/router-experts-data data/experts/code.algorithms.jsonl.
- base:
Qwen/Qwen3-4B-Instruct-2507
- rank 16, alpha 32, dropout 0.05, modules q_proj, k_proj, v_proj, o_proj
- lr 0.0002, epochs 2, max_len 1024, token budget 8192 per batch
- one example per item per recipe (direct,cot_short), under the serving system prompt; 40% of the mc items re-lettered to 5 to 10 options
- loss on the assistant turn only; held-out 5% for the val loss
Table with columns: field, value| field | value |
|---|
| target | code.algorithms |
| items | 1500 |
| recipes | direct,cot_short |
| kinds | {'code': 1500} |
| mc_padded | 0 |
| examples | 3000 |
| encoded | 2998 |
| dropped_too_long | 2 |
| train_rows | 2848 |
| val_rows | 150 |
| train_batches | 97 |
| train_tokens | 767364 |
| supervised_tokens | 387246 |
| steps | 194 |
| train_loss | 0.14857557573422944 |
| val_loss_before | 0.9747351106604415 |
| val_loss | 0.12078290439546512 |
| seconds | 269.24 |
| trainable_params | 11796480 |