README.md

metadata

language:
  - en
license: llama3.1
datasets:
  - BAAI/Infinity-Instruct
model-index:
  - name: Infinity-Instruct-7M-Gen-Llama3_1-8B
    results:
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: IFEval (0-Shot)
          type: HuggingFaceH4/ifeval
          args:
            num_few_shot: 0
        metrics:
          - type: inst_level_strict_acc and prompt_level_strict_acc
            value: 61.32
            name: strict accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=BAAI/Infinity-Instruct-7M-Gen-Llama3_1-8B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: BBH (3-Shot)
          type: BBH
          args:
            num_few_shot: 3
        metrics:
          - type: acc_norm
            value: 30.89
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=BAAI/Infinity-Instruct-7M-Gen-Llama3_1-8B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MATH Lvl 5 (4-Shot)
          type: hendrycks/competition_math
          args:
            num_few_shot: 4
        metrics:
          - type: exact_match
            value: 9.74
            name: exact match
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=BAAI/Infinity-Instruct-7M-Gen-Llama3_1-8B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: GPQA (0-shot)
          type: Idavidrein/gpqa
          args:
            num_few_shot: 0
        metrics:
          - type: acc_norm
            value: 5.7
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=BAAI/Infinity-Instruct-7M-Gen-Llama3_1-8B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MuSR (0-shot)
          type: TAUR-Lab/MuSR
          args:
            num_few_shot: 0
        metrics:
          - type: acc_norm
            value: 5.3
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=BAAI/Infinity-Instruct-7M-Gen-Llama3_1-8B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MMLU-PRO (5-shot)
          type: TIGER-Lab/MMLU-Pro
          config: main
          split: test
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 24.71
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=BAAI/Infinity-Instruct-7M-Gen-Llama3_1-8B
          name: Open LLM Leaderboard

Infinity Instruct

Beijing Academy of Artificial Intelligence (BAAI)
[Paper][Code][🤗] (would be released soon)

Infinity-Instruct-7M-Gen-Llama3.1-8B is an opensource supervised instruction tuning model without reinforcement learning from human feedback (RLHF). This model is just finetuned on Infinity-Instruct-7M and Infinity-Instruct-Gen and showing favorable results on AlpacaEval 2.0 compared to GPT4.

News

• 🔥🔥🔥[2024/08/02] We release the model weights of InfInstruct-Llama3.1-70B Gen, InfInstruct-Llama3.1-8B Gen, InfInstruct-Mistral-7B Gen.

• 🔥🔥🔥[2024/08/02] We release the 7M foundational dataset Infinity-Instruct-7M.

• 🔥🔥🔥[2024/07/09] We release the model weights of InfInstruct-Mistral-7B 0625, InfInstruct-Qwen2-7B 0625, InfInstruct-Llama3-8B 0625, InfInstruct-Llama3-8B 0625, and InfInstruct-Yi-1.5-9B 0625.

• 🔥🔥🔥[2024/07/09] We release the chat dataset Infinity-Instruct-0625, it is a upgraded version of the Infinity-Instruct-0613.

• 🔥🔥🔥[2024/06/28] We release the model weight of InfInstruct-Llama3-8B 0613. It shows favorable results on AlpacaEval 2.0 compared to GPT4-0613 without RLHF.

• 🔥🔥🔥[2024/06/21] We release the model weight of InfInstruct-Mistral-7B 0613. It shows favorable results on AlpacaEval 2.0 compared to Mixtral 8x7B v0.1, Gemini Pro, and GPT-3.5 without RLHF.

• 🔥🔥🔥[2024/06/13] We share the intermediate result of our data construction process (corresponding to the InfInstruct-3M in the table below). Our ongoing efforts focus on risk assessment and data generation. The finalized version with 10 million instructions is scheduled for release in late June.

Training Details

Infinity-Instruct-7M-Gen-Llama3.1-8B is tuned on Million-level instruction dataset Infinity-Instruct. First, we apply the foundational dataset Infinity-Instruct-7M to improve the foundational ability (math & code) of Llama3-8B, and get the foundational instruct model Infinity-Instruct-7M-Llama3.1-8B. Then we finetune the Infinity-Instruct-7M-Llama3.1-8B to get the stronger chat model Infinity-Instruct-7M-Gen-Llama3.1-8B. Here is the training hyperparamers.

epoch: 3
lr: 5e-6
min_lr: 0
lr_warmup_steps: 40
lr_decay_style: cosine
weight_decay: 0.0
adam_beta1: 0.9
adam_beta2: 0.95
global_batch_size: 528
clip_grad: 1.0

Thanks to FlagScale, we could concatenate multiple training samples to remove padding token and apply diverse acceleration techniques to the traning procudure. It effectively reduces our training costs. We will release our code in the near future!

Benchmark

Model	MT-Bench	AlpacaEval2.0	Arena-hard
GPT-4-0314	9.0	35.3	50.0
GPT-4-0613	9.2	30.2	37.9
GPT-4-1106	9.3	30.2	--
Llama-3-8B-Instruct	9.0	34.4	46.6
Llama-3.1-8B-Instruct	--	20.9	20.6
InfInstruct-7M-Llama-3.1-8B	8.2	33.9	30.4

*denote the model is finetuned without reinforcement learning from human feedback (RLHF).

How to use

Infinity-Instruct-7M-Gen-Llama3.1-8B adopt the same chat template of Llama3-8B-instruct:

<|begin_of_text|><|start_header_id|>user<|end_header_id|>

How are you?<|eot_id|><|start_header_id|>assistant<|end_header_id|>

Hi!<|eot_id|><|start_header_id|>user<|end_header_id|>

How are you?<|eot_id|><|start_header_id|>assistant<|end_header_id|>

To apply this model and template in conversation scenarios, you can refer to the following code:

from transformers import AutoModelForCausalLM, AutoTokenizer, LogitsProcessorList
import torch
device = "cuda" # the device to load the model onto

model = AutoModelForCausalLM.from_pretrained("BAAI/Infinity-Instruct-7M-Gen-Llama3_1-8B",
    torch_dtype=torch.bfloat16,
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained("BAAI/Infinity-Instruct-7M-Gen-Llama3_1-8B")

prompt = "Give me a short introduction to large language model."
messages = [
    {"role": "user", "content": prompt}
]

text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)
model_inputs = tokenizer([text], return_tensors="pt").to(device)

logits_processor = LogitsProcessorList(
            [
                MinLengthLogitsProcessor(1, eos_token_id=tokenizer.eos_token_id),
                TemperatureLogitsWarper(0.7),
            ]
 )
 
generated_ids = model.generate(
    model_inputs.input_ids,
    logits_processor=logits_processor,
    max_new_tokens=512
)

generated_ids = [
    output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
]

response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
print(response)

Disclaimer

The resources, including code, data, and model weights, associated with this project are restricted for academic research purposes only and cannot be used for commercial purposes. The content produced by any version of Infinity Instruct is influenced by uncontrollable variables such as randomness, and therefore, the accuracy of the output cannot be guaranteed by this project. This project does not accept any legal liability for the content of the model output, nor does it assume responsibility for any losses incurred due to the use of associated resources and output results.

Citation

Our paper, detailing the development and features of the Infinity Instruct dataset and finetuned models, will be released soon on arXiv. Stay tuned!

@article{InfinityInstruct2024,
  title={Infinity Instruct},
  author={Beijing Academy of Artificial Intelligence (BAAI)},
  journal={arXiv preprint arXiv:2406.XXXX},
  year={2024}
}

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	22.94
IFEval (0-Shot)	61.32
BBH (3-Shot)	30.89
MATH Lvl 5 (4-Shot)	9.74
GPQA (0-shot)	5.70
MuSR (0-shot)	5.30
MMLU-PRO (5-shot)	24.71