[NeurIPS 2024] How do Large Language Models Handle Multilingualism?

This repository contains code for the paper "How do Large Language Models Handle Multilingualism?". Below is our hypothesized multilingual workflow MWork.

Layer Embedding Decoding

Installation

The package can be installed by running the following command at the root of this repository:

conda create -n SeaExam python=3.9
conda activate Layers
pip install -r requirement.txt

Running

Figure 1 in the paper is obtained by decoding the embedding by decoding the hidden embeddings of each layer to tokens within the LLM’s vocabulary. Specifically, codes are in ./layer, where the test_layer.py is the function while for other files you need to change transformers package. Use the following command to run the experiment.

cd layers
python test_layer.py

Parameters

1. Model layers

   Note that you need to change candidate_premature_layers and candidate_langs in test_layer.py; early_exit_layers of ./transformers/models/{model_name}/modeling_{model_name}.py and early_exit_layers in ./generation/utils.py, to corresponding layer index list. For example, for llama3, candidate_premature_layers must be set as

   candidate_premature_layers = [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]

2. Top-k tokens

   To keep the generalization of decoded tokens, we normally decode top-k tokens on each position, which is set as follows in /generation/utils.py

   topk_values, topk_indices = torch.topk(logit, 100, dim=-1)

Neuron Detection (PLND)

We provide codes for detecting neurons in Llama, Mistral and Gemma.

Installation

The package can be installed by running the following command at the root of this repository:

conda create -n SeaExam python=3.9
conda activate Neuron
pip install -r requirement.txt

Running

Detect corpus is stored in ./neuron_detection/corpus_all, while similar to previous settings, we need to change transformers package. When detecting, we need to define the language and number of documents used to detect. Detected neurons will be stored in folder ./output_neurons.

cd /neuron_detection
python neuron_detection.py english 1000

Parameters

Number of Top-k neurons in each layer

top_number_attn = 1000
top_number_ffn = 2000

Neuron Deactivation

We provide codes for detecting neurons in Llama, Mistral and Gemma.

Installation

The package can be installed by running the following command at the root of this repository:

conda create -n SeaExam python=3.9
conda activate Deactivate
pip install -r requirement.txt

Running

We need to change transformers package.

cd /neuron_deactivate
python test_mistral_gsm.py {language} {understanding layer} {generation layer} {attn deact_number} {ffn deact_number} {whether under_attn} {whether reason_attn} {whether gen_attn} {whether under_ffn} {whether reason_ffn} {whether gen_ffn}

Neuron Specific Enhancement

Neuron specific tuning code is the same for all models.

Installation

The package can be installed by running the following command at the root of this repository:

conda create -n SeaExam python=3.9
conda activate Enhance
pip install -r requirement.txt

Running

We need to change transformers package.

cd /neuron_enhancement
python train_neuron.py

Parameters

Note that attn_k and attn_v needs to be divided by kv_repeat. index_keys requires fitting to model you want to train and number of understanding layer and generation layer needs to be changed correspondingly.

index_keys = [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]         

index_keys_under = [i for i in range(8)]
index_keys_gen = [31-i for i in range(4)]

attn_k = {key: {num//4 for num in value} for key, value in attn_k.items()}
attn_v = {key: {num//4 for num in value} for key, value in attn_v.items()}

Citation

If you found this repository useful, please consider

@inproceedings{zhao2024large,
  title={How do Large Language Models Handle Multilingualism?},
  author={Zhao, Yiran and Zhang, Wenxuan and Chen, Guizhen and Kawaguchi, Kenji and Bing, Lidong},
  booktitle={Advances in Neural Information Processing Systems (NeurIPS)},
  year={2024}
}