WizardLMTeam/WizardMath-7B-V1.1

Model

86

stars

19

commits

8

repos using this model

1

linked in READMEs

Jan 12, 2024

updated

mistral
pytorch
text-generation
text-generation-inference
transformers
Browse cluster: Large Language Model Training & Inference β†’

README

WizardMath: Empowering Mathematical Reasoning for Large Language Models via Reinforced Evol-Instruct (RLEIF)

🏠 Home Page

πŸ€— HF Repo β€’πŸ± Github Repo β€’ 🐦 Twitter

πŸ“ƒ [WizardLM] β€’ πŸ“ƒ [WizardCoder] β€’ πŸ“ƒ [WizardMath]

πŸ‘‹ Join our Discord

News

[12/19/2023] πŸ”₯ We released WizardMath-7B-V1.1 trained from Mistral-7B, the SOTA 7B math LLM, achieves 83.2 pass@1 on GSM8k, and 33.0 pass@1 on MATH. Use this [Demo] to chat with it.

[12/19/2023] πŸ”₯ WizardMath-7B-V1.1 outperforms ChatGPT 3.5, Gemini Pro, Mixtral MOE, and Claude Instant on GSM8K pass@1.

[12/19/2023] πŸ”₯ WizardMath-7B-V1.1 is comparable with ChatGPT 3.5, Gemini Pro, and surpasses Mixtral MOE on MATH pass@1.

ModelCheckpointPaperGSM8kMATHDemo
WizardMath-7B-V1.1πŸ€— HF LinkπŸ“ƒ [WizardMath]83.233.0[Demo]
WizardMath-70B-V1.0πŸ€— HF LinkπŸ“ƒ [WizardMath]81.622.7
WizardMath-13B-V1.0πŸ€— HF LinkπŸ“ƒ [WizardMath]63.914.0
WizardMath-7B-V1.0πŸ€— HF LinkπŸ“ƒ [WizardMath]54.910.7

[12/19/2023] Comparing WizardMath-7B-V1.1 with other open source 7B size math LLMs.

ModelGSM8k Pass@1MATH Pass@1
MPT-7B6.83.0
Llama 1-7B11.02.9
Llama 2-7B12.32.8
Yi-6b32.65.8
Mistral-7B37.89.1
Qwen-7b47.89.3
RFT-7B50.3--
MAmmoTH-7B (COT)50.510.4
WizardMath-7B-V1.054.910.7
Abel-7B-00159.713
MetaMath-7B66.519.8
Arithmo-Mistral-7B74.725.3
MetaMath-Mistral-7B77.728.2
Abel-7B-00280.429.5
WizardMath-7B-V1.183.233.0

[12/19/2023] Comparing WizardMath-7B-V1.1 with large open source (30B~70B) LLMs.

ModelGSM8k Pass@1MATH Pass@1
Llemma-34B51.525.0
Minerva-62B52.427.6
Llama 2-70B56.813.5
DeepSeek 67B63.4--
Gork 33B62.923.9
MAmmoTH-70B72.421.1
Yi-34B67.915.9
Mixtral 8x7B74.428.4
MetaMath-70B82.326.6
WizardMath-7B-V1.183.233.0

❗ Data Contamination Check:

Before model training, we carefully and rigorously checked all the training data, and used multiple deduplication methods to verify and prevent data leakage on GSM8k and MATH test set.

πŸ”₯ ❗Note for model system prompts usage:

Please use the same systems prompts strictly with us, and we do not guarantee the accuracy of the quantified versions.

Default version:

"Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n### Instruction:\n{instruction}\n\n### Response:"

CoT Version: οΌˆβ—For the simple math questions, we do NOT recommend to use the CoT prompt.οΌ‰

"Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n### Instruction:\n{instruction}\n\n### Response: Let's think step by step."

Inference WizardMath Demo Script

We provide the WizardMath inference demo code here.

Citation

Please cite the repo if you use the data, method or code in this repo.

@article{luo2023wizardmath,
  title={WizardMath: Empowering Mathematical Reasoning for Large Language Models via Reinforced Evol-Instruct},
  author={Luo, Haipeng and Sun, Qingfeng and Xu, Can and Zhao, Pu and Lou, Jianguang and Tao, Chongyang and Geng, Xiubo and Lin, Qingwei and Chen, Shifeng and Zhang, Dongmei},
  journal={arXiv preprint arXiv:2308.09583},
  year={2023}
}

Contributors

WizardLM

19 commits

WizardLMTeam/WizardMath-7B-V1.1

Model

86

stars

19

commits

8

repos using this model

1

linked in READMEs

Jan 12, 2024

updated

mistral
pytorch
text-generation
text-generation-inference
transformers
Browse cluster: Large Language Model Training & Inference β†’

README

WizardMath: Empowering Mathematical Reasoning for Large Language Models via Reinforced Evol-Instruct (RLEIF)

🏠 Home Page

πŸ€— HF Repo β€’πŸ± Github Repo β€’ 🐦 Twitter

πŸ“ƒ [WizardLM] β€’ πŸ“ƒ [WizardCoder] β€’ πŸ“ƒ [WizardMath]

πŸ‘‹ Join our Discord

News

[12/19/2023] πŸ”₯ We released WizardMath-7B-V1.1 trained from Mistral-7B, the SOTA 7B math LLM, achieves 83.2 pass@1 on GSM8k, and 33.0 pass@1 on MATH. Use this [Demo] to chat with it.

[12/19/2023] πŸ”₯ WizardMath-7B-V1.1 outperforms ChatGPT 3.5, Gemini Pro, Mixtral MOE, and Claude Instant on GSM8K pass@1.

[12/19/2023] πŸ”₯ WizardMath-7B-V1.1 is comparable with ChatGPT 3.5, Gemini Pro, and surpasses Mixtral MOE on MATH pass@1.

ModelCheckpointPaperGSM8kMATHDemo
WizardMath-7B-V1.1πŸ€— HF LinkπŸ“ƒ [WizardMath]83.233.0[Demo]
WizardMath-70B-V1.0πŸ€— HF LinkπŸ“ƒ [WizardMath]81.622.7
WizardMath-13B-V1.0πŸ€— HF LinkπŸ“ƒ [WizardMath]63.914.0
WizardMath-7B-V1.0πŸ€— HF LinkπŸ“ƒ [WizardMath]54.910.7

[12/19/2023] Comparing WizardMath-7B-V1.1 with other open source 7B size math LLMs.

ModelGSM8k Pass@1MATH Pass@1
MPT-7B6.83.0
Llama 1-7B11.02.9
Llama 2-7B12.32.8
Yi-6b32.65.8
Mistral-7B37.89.1
Qwen-7b47.89.3
RFT-7B50.3--
MAmmoTH-7B (COT)50.510.4
WizardMath-7B-V1.054.910.7
Abel-7B-00159.713
MetaMath-7B66.519.8
Arithmo-Mistral-7B74.725.3
MetaMath-Mistral-7B77.728.2
Abel-7B-00280.429.5
WizardMath-7B-V1.183.233.0

[12/19/2023] Comparing WizardMath-7B-V1.1 with large open source (30B~70B) LLMs.

ModelGSM8k Pass@1MATH Pass@1
Llemma-34B51.525.0
Minerva-62B52.427.6
Llama 2-70B56.813.5
DeepSeek 67B63.4--
Gork 33B62.923.9
MAmmoTH-70B72.421.1
Yi-34B67.915.9
Mixtral 8x7B74.428.4
MetaMath-70B82.326.6
WizardMath-7B-V1.183.233.0

❗ Data Contamination Check:

Before model training, we carefully and rigorously checked all the training data, and used multiple deduplication methods to verify and prevent data leakage on GSM8k and MATH test set.

πŸ”₯ ❗Note for model system prompts usage:

Please use the same systems prompts strictly with us, and we do not guarantee the accuracy of the quantified versions.

Default version:

"Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n### Instruction:\n{instruction}\n\n### Response:"

CoT Version: οΌˆβ—For the simple math questions, we do NOT recommend to use the CoT prompt.οΌ‰

"Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n### Instruction:\n{instruction}\n\n### Response: Let's think step by step."

Inference WizardMath Demo Script

We provide the WizardMath inference demo code here.

Citation

Please cite the repo if you use the data, method or code in this repo.

@article{luo2023wizardmath,
  title={WizardMath: Empowering Mathematical Reasoning for Large Language Models via Reinforced Evol-Instruct},
  author={Luo, Haipeng and Sun, Qingfeng and Xu, Can and Zhao, Pu and Lou, Jianguang and Tao, Chongyang and Geng, Xiubo and Lin, Qingwei and Chen, Shifeng and Zhang, Dongmei},
  journal={arXiv preprint arXiv:2308.09583},
  year={2023}
}

Contributors

WizardLM

19 commits