Tune default SFT and LoRA hyperparameters

SKILL.md

@@ -89,11 +89,29 @@ The default experiment uses:
 - 1 epoch
 - LoRA rank 32 for LoRA runs
 - bf16 full fine-tuning for full runs
+- full SFT learning rate `1e-5`
+- LoRA learning rate `5e-5`
+- warmup ratio `0.1`
+- explicit cosine LR scheduler via `--lr_scheduler_type cosine`
 - `max_length=262144`
+- conservative per-device train batch size `1`
 - checkpoint save every 1000 steps
 - validation every 1000 steps
 - TensorBoard logging under `runs/`
 
+Batch size is intentionally conservative because B300/g0049 has ~275GB per GPU but the default context length is 262144 tokens. Increase batch size from the shell only after checking memory:
+
+```bash
+export PER_DEVICE_BATCH_SIZE=2
+export GRAD_ACCUM_STEPS=2
+export LORA_PER_DEVICE_BATCH_SIZE=2
+export FULL_PER_DEVICE_BATCH_SIZE=1
+export QWEN35_9B_LORA_R32_PER_DEVICE_BATCH_SIZE=2
+export QWEN36_27B_FULL_BF16_PER_DEVICE_BATCH_SIZE=1
+```
+
+Run-specific variables have the highest precedence, then global `PER_DEVICE_BATCH_SIZE` / `GRAD_ACCUM_STEPS`, then train-type defaults, then the safe default of 1.
+
 Override model IDs or paths with:
 
 ```bash