BiLLa: A Bilingual LLaMA with Enhanced Reasoning Ability

BiLLa is an open-source reasoning-enhanced bilingual LLaMA model. The main features are:

• Greatly improve the ability of Chinese language modeling, and minimize the damage to the original English ability of LLaMA;
• During the training, more task data is added with ChatGPT-generated analysis;
• Full-parameter optimization for better performance.

Github: https://github.com/Neutralzz/BiLLa

Note: Due to LLaMA's license, the model weights in this hub cannot be used directly. The weight of word embedding is the sum of the weights of the trained model and the original LLaMA, so as to ensure that developers with LLaMA original model accessibility can convert the model released by this hub into a usable one.

Usage

First, you can revert the model weights by this script:

python3 embedding_convert.py \
    --model_dir /path_to_BiLLa/BiLLa-7B-SFT \
    --meta_llama_pth_file /path_to_LLaMA/llama-7b/consolidated.00.pth

Then, you can run this model as follows:

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
model_path = "/path_to_BiLLa/BiLLa-7B-SFT"
tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=False)
model = AutoModelForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, torch_dtype=torch.float16).cuda()

prompt = "Human: Write a Python function that checks if a given number is even or odd.\nAssistant: "
input_ids = tokenizer([prompt]).input_ids
output_ids = model.generate(
            torch.as_tensor(input_ids).cuda(),
            do_sample=True,
            temperature=0.7,
            max_new_tokens=1024
        )
output_ids = output_ids[0][len(input_ids[0]):]

outputs = tokenizer.decode(output_ids, skip_special_tokens=True).strip()
print(outputs)

Input Format

Different from BiLLa-7B-LLM, the model input of BiLLa-7B-SFT should be formatted as follows:

Human: [Your question]
Assistant: 

Note that a space is following the Assistant: