Orion-14B-Base / README_en.md
liuyongq's picture
Update README_en.md
33d4faf verified
|
raw
history blame
18.8 kB

Orion-14B

Table of Contents

Model Introduction

  • Orion-14B series models are open-source multilingual large language models trained from scratch by OrionStarAI. The base model is trained on 2.5T multilingual corpus, including Chinese, English, Japanese, Korean, etc, and it exhibits superior performance in these languages.

  • In mainstream benchmark evaluations, the Orion-14B series models demonstrate outstanding competitiveness, significantly surpassing models of similar scales. Based on benchmark results, the Orion-14B series models are also the first to be evaluated across more than three languages in the domain of LLM. We hope that the contributions of all Orion Star colleagues establish a new benchmark for the research field of multilingual LLMs.

  • Orion-14B series models including:

    • Orion-14B-Base: A multilingual large language foundational model with 14 billion parameters, pretrained on a diverse dataset of 2.5 trillion tokens.
    • Orion-14B-Chat: A chat-model fine-tuned on a high-quality corpus aims to provide an excellence interactive experience for users in the large model community.
    • Orion-14B-LongChat: This model is optimized for long context lengths more than 200k tokens and demonstrates performance comparable to proprietary models on long context evaluation sets.
    • Orion-14B-RAG: A chat-model fine-tuned on a custom retrieval augmented generation dataset, achieving superior performance in retrieval augmented generation tasks.
    • Orion-14B-PlugIn: A chat-model specifically tailored for plugin and function calling tasks, ideal for agent-related scenarios where the LLM acts as a plugin and function call system.
    • Orion-14B-Base-Int4: A quantized base model utilizing 4-bit integer weights. It significantly reduces the model size by 70% and increases the inference speed by 30% while incurring a minimal performance loss of only 1%.
    • Orion-14B-Chat-Int4: A quantized chat model utilizing 4-bit integer weights.

Model Download

Model release and download links are provided in the table below:

Model Name HuggingFace Download Links ModelScope Download Links
⚾Orion-14B-Base Orion-14B-Base Orion-14B-Base
😛Orion-14B-Chat Orion-14B-Chat Orion-14B-Chat
📃Orion-14B-LongChat Orion-14B-LongChat Orion-14B-LongChat
🔎Orion-14B-Chat-RAG Orion-14B-Chat-RAG Orion-14B-Chat-RAG
🔌Orion-14B-Chat-Plugin Orion-14B-Chat-Plugin Orion-14B-Chat-Plugin
💼Orion-14B-Base-Int4 Orion-14B-Base-Int4 Orion-14B-Base-Int4
📦Orion-14B-Chat-Int4 Orion-14B-Chat-Int4 Orion-14B-Chat-Int4

Model Benchmarks

LLM evaluation results on examination and professional knowledge

Model C-Eval CMMLU MMLU AGIEval Gaokao BBH
LLaMA2-13B 41.4 38.4 55.0 30.9 18.2 45.6
Skywork-13B 59.1 61.4 62.7 43.6 56.1 48.3
Baichuan2-13B 59.0 61.3 59.5 37.4 45.6 49.0
QWEN-14B 71.7 70.2 67.9 51.9 62.5 53.7
InternLM-20B 58.8 59.0 62.1 44.6 45.5 52.5
Orion-14B 72.9 70.6 69.9 54.7 62.1 56.5

LLM evaluation results on language understanding and common knowledge

Model RACE-middle RACE-high HellaSwag PIQA Lambada WSC
LLaMA 2-13B 63.0 58.9 77.5 79.8 76.5 66.3
Skywork-13B 87.6 84.1 73.7 78.3 71.8 66.3
Baichuan 2-13B 68.9 67.2 70.8 78.1 74.1 66.3
QWEN-14B 93.0 90.3 80.2 79.8 71.4 66.3
InternLM-20B 86.4 83.3 78.1 80.3 71.8 68.3
Orion-14B 93.3 91.3 78.5 79.5 78.9 70.2

LLM evaluation results of OpenCompass testsets

Model Average Examination Language Knowledge Understanding Reasoning
LLaMA 2-13B 47.3 45.2 47.0 58.3 50.9 43.6
Skywork-13B 53.6 61.1 51.3 52.7 64.5 45.2
Baichuan 2-13B 49.4 51.8 47.5 48.9 58.1 44.2
QWEN-14B 62.4 71.3 52.67 56.1 68.8 60.1
InternLM-20B 59.4 62.5 55.0 60.1 67.3 54.9
Orion-14B 64.4 71.4 55.0 60.0 71.9 61.6

Comparison of LLM performances on Japanese testsets

Model Average JCQA JNLI MARC JSQD JQK XLS XWN MGSM
PLaMo-13B 52.3 56.7 42.8 95.8 70.6 71.0 8.70 70.5 2.40
WebLab-10B 50.7 66.6 53.7 82.1 62.9 56.2 10.0 72.0 2.40
ELYZA-jp-7B 48.8 71.7 25.3 86.6 70.8 64.1 2.50 62.1 7.20
StableLM-jp-7B 51.1 33.4 43.3 96.7 70.6 78.1 10.7 72.8 2.80
LLaMA 2-13B 46.3 75.0 47.6 38.8 76.1 67.7 18.1 63.2 10.4
Baichuan 2-13B 57.1 73.7 31.3 91.6 80.5 63.3 18.6 72.2 25.2
QWEN-14B 65.8 85.9 60.7 97.0 83.3 71.8 18.8 70.6 38.0
Yi-34B 67.1 83.8 61.2 95.2 86.1 78.5 27.2 69.2 35.2
Orion-14B 69.1 88.2 75.8 94.1 75.7 85.1 17.3 78.8 38.0

Comparison of LLM performances on Korean testsets. n = 0 and n = 5 stand for n-shot prompts used in the evaluation

Model Average
n=0  n=5
HellaSwag
n=0  n=5
COPA
n=0  n=5
BooIQ
n=0  n=5
SentiNeg
n=0  n=5
KoGPT 53.0    70.1 55.9    58.3 73.5    72.9 45.1    59.8 37.5    89.4
Polyglot-ko-13B 69.6    73.7 59.5    63.1 79.4    81.1 48.2    60.4 91.2    90.2
LLaMA 2-13B 46.7    63.7 41.3    44.0 59.3    63.8 34.9    73.8 51.5    73.4
Baichuan 2-13B 52.1    58.7 39.2    39.6 60.6    60.6 58.4    61.5 50.3    72.9
QWEN-14B 53.8    73.7 45.3    46.8 64.9    68.9 33.4    83.5 71.5    95.7
Yi-34B 54.2    72.1 44.6    44.7 58.0    60.6 65.9    90.2 48.3    92.9
Orion-14B 74.5    79.6 47.0    49.6 77.7    79.4 81.6    90.7 92.4    98.7

Multilingual evaluation

Model Train Lang Japanese Korean Chinese English
PLaMo-13B En,Jp 52.3 * * *
Weblab-10B En,Jp 50.7 * * *
ELYZA-jp-7B En,Jp 48.8 * * *
StableLM-jp-7B En,Jp 51.1 * * *
KoGPT-6B En,Ko * 70.1 * *
Polyglot-ko-13B En,Ko * 70.7 * *
Baichuan2-13B Multi 57.1 58.7 50.8 57.1
Qwen-14B Multi 65.8 73.7 64.5 65.4
Llama2-13B Multi 46.3 63.7 41.4 55.3
Yi-34B Multi 67.1 72.2 58.7 68.8
Orion-14B Multi 69.1 79.5 67.9 67.3

Evaluation for data contamination

Model C-Eval CMMLU MMLU Lambada HellaSwag
GPT-4 69.9 71.0 83.0 65.5 91.4
Qwen-72B 83.3 61.8 77.3 76.1 85.4
Yi-34B 81.8 82.6 76.3 73.1 82.0
Orion-14B 72.8 70.6 69.9 78.8 78.5
Orion-14B(contaminated) 92.7 82.9 85.4 78.5 85.8

Chat model standard evaluation

Model CMMLU MMLU BBH HellaSwag PIQA WSC
Baichuan2-13B-Chat 58.4 57.0 49.9 66.9 77.6 71.2
Qwen-14B-Chat 70.0 66.4 58.0 65.2 74.0 66.3
Llama2-13B-Chat 38.7 54.6 40.2 78.2 78.8 68.3
InternLM-20B-Chat 52.2 52.5 35.3 69.2 76.7 61.5
Orion-14B-Chat 63.7 61.71 49.05 76.7 78.4 71.15

Chat model subjective evaluation of MTBench

Model First-Turn Second-Turn Average
Baichuan2-13B-Chat 7.05 6.47 6.76
Qwen-14B-Chat 7.30 6.62 6.96
Llama2-13B-Chat 7.10 6.20 6.65
InternLM-20B-Chat 7.03 5.93 6.48
Orion-14B-Chat 7.68 7.07 7.37

Chat model subjective evaluation of AlignBench

Model Math. Logi. Basic. Chi. Comp. Writ. Role. Prof. Avg.
Baichuan2-13B-Chat 3.76 4.07 6.22 6.05 7.11 6.97 6.75 6.43 5.25
Qwen-14B-Chat 4.91 4.71 6.90 6.36 6.74 6.64 6.59 6.56 5.72
Llama2-13B-Chat 3.05 3.79 5.43 4.40 6.76 6.63 6.99 5.65 4.70
InternLM-20B-Chat 3.39 3.92 5.96 5.50 7.18 6.19 6.49 6.22 4.96
Orion-14B-Chat 4.00 4.24 6.18 6.57 7.16 7.36 7.16 6.99 5.51

Model Inference

Model weights, source code, and configuration needed for inference are published on Hugging Face, and the download link is available in the table at the beginning of this document. We demonstrate various inference methods here, and the program will automatically download the necessary resources from Hugging Face.

Python Code

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
from transformers.generation.utils import GenerationConfig

tokenizer = AutoTokenizer.from_pretrained("OrionStarAI/Orion-14B", use_fast=False, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("OrionStarAI/Orion-14B", device_map="auto",
                                             torch_dtype=torch.bfloat16, trust_remote_code=True)

model.generation_config = GenerationConfig.from_pretrained("OrionStarAI/Orion-14B")
messages = [{"role": "user", "content": "Hello, what is your name? "}]
response = model.chat(tokenizer, messages, streaming=False)
print(response)

# Hello, my name is ChatMax. Nice to meet you.

In the above Python code, the model is loaded with device_map='auto' to utilize all available GPUs. To specify the device, you can use something like export CUDA_VISIBLE_DEVICES=0,1 (using GPUs 0 and 1).

Command Line Tool

python cli_demo.py

This command-line tool is designed for chat scenarios, and thus, it does not support calling the base model.

Direct Script Inference

 python text_generation.py --model OrionStarAI/Orion-14B --tokenizer OrionStarAI/Orion-14B --prompt 你好,你叫什么名字

Example Output

Casual Chat

User: Hello, what is your name?
Orion-14B: Hello, my name is ChatMax. Nice to meet you.
User: What functions do you have?
Orion-14B: I am a versatile AI assistant. I can answer your questions, chat with you, help you look up information, provide life advice, and more.
User: Tell me a joke.
Orion-14B: Why is sausage always unhappy? Because it has no friends and can only be sandwiched between bread.

Mathematical Problem

User: In a cage, there are chickens and rabbits with a total of 26 heads and 68 feet. So, how many chickens and how many rabbits are there?
Orion-14B: Let the number of chickens be x, and the number of rabbits be y.
According to the problem, we can derive the following two equations:
x + y = 26 (Total number of heads)
2x + 4y = 68 (Total number of feet)
Rewriting the first equation, we get y = 26 - x.
Substituting the value of y into the second equation, we get:
2x + 4(26 - x) = 68
Solving this equation, we find x = 18.
Therefore, there are 18 chickens and 26 - 18 = 8 rabbits.

Company Introduction

OrionStar is a leading global service robot solutions company, founded in September 2016. OrionStar is dedicated to using artificial intelligence technology to create the next generation of revolutionary robots, allowing people to break free from repetitive physical labor and making human work and life more intelligent and enjoyable. Through technology, OrionStar aims to make society and the world a better place.

OrionStar possesses fully self-developed end-to-end artificial intelligence technologies, such as voice interaction and visual navigation. It integrates product development capabilities and technological application capabilities. Based on the Orion robotic arm platform, it has launched products such as OrionStar AI Robot Greeting, AI Robot Greeting Mini, Lucki, Coffee Master, and established the open platform OrionOS for Orion robots. Following the philosophy of "Born for Truly Useful Robots", OrionStar empowers more people through AI technology.

Declarations, License

Declarations

We strongly urge all users not to use the Orion-14B model for any activities that may harm national or social security or violate the law. Additionally, we request users not to use the Orion-14B model for internet services without proper security review and filing. We hope all users abide by this principle to ensure that technological development takes place in a regulated and legal environment. We have done our best to ensure the compliance of the data used in the model training process. However, despite our significant efforts, unforeseen issues may still arise due to the complexity of the model and data. Therefore, if any problems arise due to the use of the Orion-14B open-source model, including but not limited to data security issues, public opinion risks, or any risks and issues arising from the model being misled, abused, disseminated, or improperly utilized, we will not assume any responsibility.

License

Community use of the Orion-14B model must comply with the Apache 2.0.

Contact Us

Email: ai@orionstar.com

WhatsApp Group: https://chat.whatsapp.com/J30ig8Dx4ja5jc0cfx2nVs