File size: 3,105 Bytes
2dd2573 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 | # LoRA Training
Best practice reference for single-node 8xH20 LoRA training with Qwen3-235B-A22B-Instruct-250718: https://github.com/modelscope/ms-swift/pull/5033.
For environment setup, please refer to the [Quick Start Guide](./Quick-start.md) of Megatron-SWIFT.
## Converting HF to Mcore
The conversion process is the same as for full-parameter training. Use the following script:
```shell
CUDA_VISIBLE_DEVICES=0 \
swift export \
--model Qwen/Qwen2.5-7B-Instruct \
--to_mcore true \
--torch_dtype bfloat16 \
--output_dir Qwen2.5-7B-Instruct-mcore \
--test_convert_precision true
```
## LoRA Training
Training Script:
```bash
# full: 2 * 70GiB 0.61s/it
# lora: 2 * 14GiB 0.45s/it
PYTORCH_CUDA_ALLOC_CONF='expandable_segments:True' \
NPROC_PER_NODE=2 \
CUDA_VISIBLE_DEVICES=0,1 \
megatron sft \
--load Qwen2.5-7B-Instruct-mcore \
--dataset 'AI-ModelScope/alpaca-gpt4-data-zh#500' \
'AI-ModelScope/alpaca-gpt4-data-en#500' \
'swift/self-cognition#500' \
--train_type lora \
--lora_rank 8 \
--lora_alpha 32 \
--target_modules all-linear \
--tensor_model_parallel_size 2 \
--sequence_parallel true \
--micro_batch_size 16 \
--global_batch_size 16 \
--recompute_granularity full \
--recompute_method uniform \
--recompute_num_layers 1 \
--finetune true \
--cross_entropy_loss_fusion true \
--lr 1e-4 \
--lr_warmup_fraction 0.05 \
--min_lr 1e-5 \
--max_epochs 1 \
--save megatron_output/Qwen2.5-7B-Instruct \
--save_interval 100 \
--max_length 2048 \
--system 'You are a helpful assistant.' \
--num_workers 4 \
--no_save_optim true \
--no_save_rng true \
--dataset_num_proc 4 \
--model_author swift \
--model_name swift-robot
```
- For LoRA training scripts of MoE models, please refer to [here](https://github.com/modelscope/ms-swift/tree/main/examples/megatron/lora).
## Converting MCore to HF
```bash
CUDA_VISIBLE_DEVICES=0 \
swift export \
--mcore_adapters megatron_output/Qwen2.5-7B-Instruct/vx-xxx \
--to_hf true \
--torch_dtype bfloat16 \
--output_dir megatron_output/Qwen2.5-7B-Instruct/vx-xxx-hf \
--test_convert_precision true
```
- Note: The `mcore_adapters` folder contains an `args.json` file. During the conversion process, parameters related to `mcore_model` and LoRA will be loaded from this file. The system will then perform a merge-lora operation between the `mcore_model` and `mcore_adapters` to obtain the complete model weights, and finally convert them into HuggingFace (HF) format. (Conversion of LoRA incremental weights is not supported for now)
## Merge-LoRA
If you only want to merge the LoRA weights without converting them to Hugging Face format, for subsequent DPO training, you can use the following script:
```shell
CUDA_VISIBLE_DEVICES=0 \
swift export \
--mcore_adapters megatron_output/Qwen2.5-7B-Instruct/vx-xxx \
--to_mcore true \
--torch_dtype bfloat16 \
--output_dir megatron_output/Qwen2.5-7B-Instruct/vx-xxx-mcore \
--test_convert_precision true
```
|