ayo/ipex-llm
1
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

[LLM] Add llama3 alpaca qlora example (#11011)

Browse source 
* Add llama3 finetune example based on alpaca qlora example
This commit is contained in:
Ziteng Zhang 2024-05-15 09:17:32 +08:00 committed by GitHub
parent 0a732bebe7
commit 7d3791c819
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 47 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

18
python/llm/example/GPU/LLM-Finetuning/QLoRA/alpaca-qlora/README.md
View file

@ -5,6 +5,7 @@ This example ports [Alpaca-LoRA](https://github.com/tloen/alpaca-lora/tree/main)
> Note: You could also refer to [simple QLoRA example](../simple-example/) to try related usage. > Note: You could also refer to [simple QLoRA example](../simple-example/) to try related usage.
### 0. Requirements ### 0. Requirements
To run this example with IPEX-LLM on Intel GPUs, we have some recommended requirements for your machine, please refer to [here](../../../README.md#requirements) for more information. To run this example with IPEX-LLM on Intel GPUs, we have some recommended requirements for your machine, please refer to [here](../../../README.md#requirements) for more information.
### 1. Install ### 1. Install
@ -26,6 +27,7 @@ pip install git+https://github.com/intel/intel-extension-for-deepspeed.git@ec332
``` ```
### 2. Configures OneAPI environment variables ### 2. Configures OneAPI environment variables
```bash ```bash
source /opt/intel/oneapi/setvars.sh source /opt/intel/oneapi/setvars.sh
``` ```
@ -129,8 +131,21 @@ bash qlora_finetune_llama2_70b_pvc_1550_4_card.sh
</details> </details>
<details>
<summary> Show LLaMA3-8B examples </summary>
##### Finetuning LLaMA3-8B on single Arc A770
```bash
bash qlora_finetune_llama3_8b_arc_1_card.sh
```
</details>
### 4. (Optional) Resume Training ### 4. (Optional) Resume Training
If you fail to complete the whole finetuning process, it is suggested to resume training from a previously saved checkpoint by specifying `resume_from_checkpoint` to the local checkpoint folder as following:** If you fail to complete the whole finetuning process, it is suggested to resume training from a previously saved checkpoint by specifying `resume_from_checkpoint` to the local checkpoint folder as following:**
```bash ```bash
python ./alpaca_qlora_finetuning.py \ python ./alpaca_qlora_finetuning.py \
--base_model "meta-llama/Llama-2-7b-hf" \ --base_model "meta-llama/Llama-2-7b-hf" \
@ -140,6 +155,7 @@ python ./alpaca_qlora_finetuning.py \
``` ```
### 5. Sample Output ### 5. Sample Output
```log ```log
{'loss': 1.9231, 'learning_rate': 2.9999945367033285e-05, 'epoch': 0.0} {'loss': 1.9231, 'learning_rate': 2.9999945367033285e-05, 'epoch': 0.0}
{'loss': 1.8622, 'learning_rate': 2.9999781468531096e-05, 'epoch': 0.01} {'loss': 1.8622, 'learning_rate': 2.9999781468531096e-05, 'epoch': 0.01}
@ -153,6 +169,7 @@ python ./alpaca_qlora_finetuning.py \
``` ```
### 6. Merge the adapter into the original model ### 6. Merge the adapter into the original model
``` ```
python ./export_merged_model.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --adapter_path ./outputs/checkpoint-200 --output_path ./outputs/checkpoint-200-merged python ./export_merged_model.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --adapter_path ./outputs/checkpoint-200 --output_path ./outputs/checkpoint-200-merged
``` ```
@ -160,4 +177,5 @@ python ./export_merged_model.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --
Then you can use `./outputs/checkpoint-200-merged` as a normal huggingface transformer model to do inference. Then you can use `./outputs/checkpoint-200-merged` as a normal huggingface transformer model to do inference.
### 7. Troubleshooting ### 7. Troubleshooting
Please refer to [here](../../README.md#troubleshooting) for solutions of common issues during finetuning. Please refer to [here](../../README.md#troubleshooting) for solutions of common issues during finetuning.

21
python/llm/example/GPU/LLM-Finetuning/QLoRA/alpaca-qlora/qlora_finetune_llama3_8b_arc_1_card.sh Normal file
View file

@ -0,0 +1,21 @@
# Copyright 2016 The BigDL Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
# You could also specify `--base_model` to the local path of the huggingface model checkpoint folder and `--data_path` to the local path of the dataset JSON file
python ./alpaca_qlora_finetuning.py \
--base_model "meta-llama/Meta-Llama-3-8B-Instruct" \
--data_path "yahma/alpaca-cleaned" \
--output_dir "./ipex-llm-qlora-alpaca"