Our weights for the instruction tuning model is uploading here
TALLRec: An Effective and Efficient Tuning Framework to Align Large Language Model with Recommendation is available at https://arxiv.org/abs/2305.00447.
Wrongly delete the line in evaluate.py by mistake, now it has been updated
We introduce a novel framework (TALLRec) that enables the efficient and effective adaptation of LLMs to recommendation tasks.
| movie | book | |||||
|---|---|---|---|---|---|---|
| Few-shot | 16 | 64 | 256 | 16 | 64 | 256 |
| GRU | 49.07 | 49.87 | 52.89 | 48.95 | 49.64 | 49.86 |
| Caser | 49.68 | 51.06 | 54.20 | 49.84 | 49.72 | 49.57 |
| SASRec | 50.43 | 50.48 | 52.25 | 49.48 | 50.06 | 50.20 |
| DROS | 50.76 | 51.54 | 54.07 | 49.28 | 49.13 | 49.13 |
| GRU-BERT | 50.85 | 51.65 | 53.44 | 50.07 | 49.64 | 49.79 |
| DROS-BERT | 50.21 | 51.71 | 53.94 | 50.07 | 48.98 | 50.20 |
| TALLRec (ours) | 67.24 | 67.48 | 71.98 | 56.36 | 60.39 | 64.38 |
Table 1. we shown the AUC results of the baseline models and our frameworks on movie and book scenarios.
Train TALLRec base on LLaMA7B:
bash ./shell/instruct_7B.sh gpu_id random_seed
If you want to run it under your environment, you need to make changes to the sh file:
- output_dir: Model save path，we will automatically add the seed and the sample to the end of the path for each experiments.
- base_model: LLaMA parameter weight path in Hugginface format
- train_data: Training data path such as "./data/movie/train.json" for movie dataset.
- val_data: Validation data set path such as "./data/movie/valid.json" for movie dataset.
- instruction_model: The LoRA weights after the instruction tuning, for example lora weight from alpaca-lora.
After training, you need to evluate the test result on the best model evaluated by the validation set.
bash ./shell/evaluate.sh gpu_id output_dir
If you want to run it under your environment, you need to make changes to the sh file:
- base_model: LLaMA parameter weight path in Hugginface format
- test_data: Test data set path such as "./data/movie/test.json" for movie dataset.
Note that we will automatically detect all the different seed and sample files in the output_dir directory, and then integrate these results into the output_dir.json file.
Our project is developed based on the Alpaca_lora repo, thanks for their contributions.
For "Environment setting sharing for CUDA 12.0", please see here.