--- title: sirl-qwen2-5-7b-copt canonical_url: "https://www.modelscope.cn/models/oneday88/sirl-qwen2-5-7b-copt" md_url: "https://www.modelscope.cn/models/oneday88/sirl-qwen2-5-7b-copt.md" repository: oneday88/sirl-qwen2-5-7b-copt chinese_name: "SIRL--COPT--杉数科技" last_updated: 2025-09-09 license: "Apache License 2.0" model_type: - qwen2 architectures: - Qwen2ForCausalLM parameters: 7.6B tensor_type: - BF16 library_name: - transformer - safetensors - pytorch frameworks: - pytorch inference_backends: - "deploy_task text/emb" - "lmdeploy 0.9.1" - "lmdeploy_turbomind 0.9.1" - "sglang 0.5.2" - "vllm 0.9.2" downloads: 52 stars: 0 --- # sirl-qwen2-5-7b-copt > sirl-qwen2-5-7b-copt - oneday88 在 ModelScope 开源的模型。Overview & Examples We introduce SIRL (Solver-Informed Reinforcement Learning), a novel reasoning paradigm that integrates solver feedback with reinforcement learning to train large language models (LLMs)… oneday88/sirl-qwen2-5-7b-copt 是 ModelScope 魔搭社区上的 7.6B 参数机器学习模型，采用 Apache License 2.0 许可，可用 deploy_task text/emb、lmdeploy 0.9.1、lmdeploy_turbomind 0.9.1 部署。 - **Repository**: oneday88/sirl-qwen2-5-7b-copt - **License**: Apache License 2.0 - **Parameters**: 7.6B - **Inference backends**: deploy_task text/emb, lmdeploy 0.9.1, lmdeploy_turbomind 0.9.1, sglang 0.5.2, vllm 0.9.2 - **Downloads**: 52 - **Stars**: 0 - **Last updated**: 2025-09-09 Source: https://www.modelscope.cn/models/oneday88/sirl-qwen2-5-7b-copt --- ## Overview & Examples We introduce **SIRL (Solver-Informed Reinforcement Learning)**, a novel reasoning paradigm that integrates solver feedback with reinforcement learning to train large language models (LLMs) for optimization modeling. This approach represents the first application of Reinforcement Learning with Verifiable Reward (RLVR) in the domain of optimization modeling, enabling LLMs to generate accurate mathematical formulations and code generations from natural language descriptions. SIRL leverages solver outputs to iteratively refine model performance. Our SIRL-Qwen2.5-32B model surpasses the performance of DeepSeek-V3 and OpenAI-O3 on optimization modeling benchmarks,demonstrating the effectiveness of our approach. Currently, we offer LLM model checkpoints that seamlessly integrate with both Gurobi and COPT optimization solver. COPT (Cardinal Optimizer) is a mathematical optimization solver for large-scale optimization problems developed by Cardinal Operations, and it includes high-performance solvers for LP, MIP, NLP and so on. To explore its full functionalities or to request a trial, please visit the official website: www.shanshu.ai/copt. ## Model Release The checkpoints of [SIRL-Qwen2.5-7B-Gurobi](https://huggingface.co/chenyitian-shanshu/SIRL), [SIRL-Qwen2.5-7B-COPT](https://huggingface.co/chenyitian-shanshu/SIRL/tree/main/Copt) and [SIRL-Qwen2.5-32B-Gurobi](https://huggingface.co/chenyitian-shanshu/SIRL-Gurobi32B) are avaiable on Hugging Face and Model Scope. Looking ahead, we aim to develop our next-generation LLM models to tackle a broader range of general optimization and mathematical tasks. | Solver Type | Hugging Face | ModelScope | |---------------------|---------------- | ---| | Gurobi-7B | [SIRL-Qwen2.5-7B-Gurobi](https://huggingface.co/chenyitian-shanshu/SIRL-Gurobi) | [SIRL-Qwen2.5-7B-Gurobi](https://modelscope.cn/models/oneday88/SIRL-7B) | | Gurobi-32B | [SIRL-Qwen2.5-32B-Gurobi](https://huggingface.co/chenyitian-shanshu/SIRL-Gurobi32B) | [SIRL-Qwen2.5-32B-Gurobi](https://modelscope.cn/models/oneday88/sirl-qwen2-5-32b-gurobi) | | COPT-7B | [SIRL-Qwen2.5-7B-COPT](https://huggingface.co/chenyitian-shanshu/SIRL-COPT) | [SIRL-Qwen2.5-7B-COPT](https://modelscope.cn/models/oneday88/sirl-qwen2-5-7b-copt) | ## Performance We evaluated the performance of the proposed SIRL framework on four benchmarks: NL4OPT, MAMO, IndustryOR and OptMATH. Performance is assessed based on the pass@1 accuracy(acc). Following the rigorous evaluation protocol proposed by OptMATH, a solution is considered valid if the relative error is less than 1e-6. The performance metrics for [SIRL](https://huggingface.co/chenyitian-shanshu/SIRL) are as follows. The highest results are highlighted in bold. | Types | Models | NL4OPT | MAMO Easy fixed | MAMO Complex fixed | IndustryOR | OptMATH_166 | OptiBench | Macro AVG | |---------------|-------------------|--------|-----------|--------------|------------|---------|-----------|-----------| | Baseline | GPT-4 | 89.0%* | 87.3%* | 49.3%* | 33.0%* | 16.6%* | 68.6%* | 57.4%* | | | Deepseek-V3 | 95.9%* | 88.3%* | 50.2% | 37.0%* | 44.0% | **71.6%*** | 64.5%* | | | DeepSeek-R1 | 82.4% | 87.2% | **67.9%** | **45.0%** | 40.4% | 66.4% | 61.9% | | | OpenAI-O3 | 69.4% | 77.1% | 51.2% | 44.0% | 44.0% | 58.6% | 57.38% | | Agent-based | OptiMUS | 78.8%* | 77.0%* | 43.6%* | 31.0%* | 20.2%* | 45.8%* | 49.4%* | | Offline-learning | ORLM-LLaMA-3-8B | 85.7%* | 82.3%* | 37.4%* | 24.0%* | 2.6%* | 51.1%* | 47.2%* | | | LLMOpt-Qwen2.5-14B | 80.3%* | 89.5%* | 44.1%* | 29.0%* | 12.5%* | 53.8%* | 51.1%* | | | OptMATH-Qwen2.5-7B | 94.7%* | 86.5%* | 40.8% | 20.0%* | 24.4%* | 57.9%* | 55.8%* | | | OptMATH-Qwen2.5-32B | 95.9%| 89.9%| 54.1%| 31.0% |34.7% |66.1% |62.0% | | Gurobi-7B | SIRL-Qwen2.5-7B-Gurobi | 96.3%* | 91.7% | 51.7% | 33.0% | 30.5% | 58.0% | 60.2% | | Gruobi-32B | SIRL-Qwen2.5-32B-Gurobi | **98.0%** | **94.6%**| 61.1% |42.0% |**45.8%** |67.4% |**68.2%** | | COPT-7B | SIRL-Qwen2.5-7B-COPT| 95.1% | 92.1% | 53.1% | 31.0% | 29.5% | 58.3% | 58.9%| *Note:* Values marked with "*" are from original or reproduced papers with the criterion: relative error < 10⁻⁶. * The code to reproduce the results of Gurobi version can be found in our [Jupyter Notebook for Gurobi](https://github.com/Cardinal-Operations/SIRL/blob/main/reproduce_gurobi.ipynb). * The code to reproduce the results of COPT version can be found in our [Jupyter Notebook for COPT](https://github.com/Cardinal-Operations/SIRL/blob/main/reproduce_copt.ipynb). ## Inference ### Setup To get started, clone SIRL and install the required packages: ```shell pip install -r requirements.txt ``` Make sure that you have already apply for the license of solvers such as Gurobi or COPT. We recommend using the following prompt template which can be found in [rule_prompt_utils.py](https://github.com/Cardinal-Operations/SIRL/blob/main/rule_prompt_utils.py). Please replace the {question} with any natural language OR question. ### Quick start Below is a simple example for model inference: ```python from transformers import AutoTokenizer from rule_prompt_utils import gurobi_prompt_temp,copt_prompt_temp from utils import extract_code_block, extract_obj from vllm import SamplingParams, LLM from langchain.prompts import PromptTemplate import subprocess sovler_name = 'copt' # solver_name = 'copt' # Load model and parameters for Gurobi model = LLM("chenyitian-shanshu/SIRL-copt", tensor_parallel_size=1, trust_remote_code=True) tokenizer = AutoTokenizer.from_pretrained("chenyitian-shanshu/SIRL-copt") sampling_params = SamplingParams( n=1, temperature=0.5, top_p=0.9, max_tokens=8192, repetition_penalty=1.02 ) # Load model and parameters for COPT #model = LLM("chenyitian-shanshu/SIRL-COPT", # tensor_parallel_size=1, # trust_remote_code=True) #tokenizer = AutoTokenizer.from_pretrained("chenyitian-shanshu/SIRL-COPT") #sampling_params = SamplingParams( # n=1, # temperature=0.5, # top_p=0.9, # max_tokens=16384, # repetition_penalty=1.02 # ) # Load question. Here is just an example. Users can replace this with datasets they want to test question = "An industrial tire company delivers large tires for equipment to remote engineering sites either by cargo planes or ultrawide trucks. Each cargo plane can transport 10 tires per trip and costs $1000. Each ultrawide truck can transport 6 tires per trip and costs $700. The company needs to transport at least 200 tires and has available $22000. Because most remote sites don't have proper airports, the number of plane trips cannot exceed the number of ultrawide truck trips. How many trips of each should be done to minimize the total number of trips?" # Load prompt templete for Gurobi zeroshot_prompt_system = PromptTemplate.from_template(gurobi_prompt_temp['system']) zeroshot_prompt_user = PromptTemplate.from_template(gurobi_prompt_temp['user']) # Load prompt template for COPT #zeroshot_prompt_system = PromptTemplate.from_template(copt_prompt_temp['system']) #zeroshot_prompt_user = PromptTemplate.from_template(copt_prompt_temp['user']) prompt =[{"role": "system", "content": zeroshot_prompt_system.format().strip() }, {"role": "user", "content": zeroshot_prompt_user.format(question=question).strip() }] # Generate Response text = tokenizer.apply_chat_template(prompt, tokenize=False, add_generation_prompt=True) response = model.generate(text,sampling_params) response_text = response[0].outputs[0].text code_snippet = extract_code_block(response_text,solver_name) result = subprocess.run(['python3', '-c', code_snippet], capture_output=True, text=True, timeout=100) obj = extract_obj(result.stdout,solver_name) print(response_text) print('optimal value is', obj) ``` ## Test Dataset We evaluate the performance of our trained model on multiple datasets which include NL4OPT, MAMO, IndustryOR, OptMATH. Minor errors exist within these testing datasets. To address this, we rigorously reviewed and corrected the test sets of these benchmarks, updating the questions and corresponding answers to ensure the integrity of our evaluation, with a specific focus on the NL4OPT and IndustryOR dataset. The datasets are available at [https://github.com/Cardinal-Operations/SIRL/tree/main/test_data](https://github.com/Cardinal-Operations/SIRL/tree/main/test_data) and hugging face dataset: https://huggingface.co/datasets/chenyitian-shanshu/ORLMBenchmark ### Data Structure Each dataset is organized in a `jsonl` file, with each line containing an independent data entry. Each entry includes: - `en_question`: A string description of the optimization problem. - `en_answer`: The ground truth objective function value (float). The answers of infeasible problems are "No Best Solution" or "-99999" An example from NL4OPT: ```json { "en_question": "A company needs to minimize shipping costs across 5 warehouses with varying demands...", "en_answer": 1250.50, } ``` ## Citation If you find SILR useful or relevant to your research, please consider citing our paper: ```bibtex @article{chen2025solver, title={Solver-Informed RL: Grounding Large Language Models for Authentic Optimization Modeling}, author={Chen, Yitian and Xia, Jingfan and Shao, Siyu and Ge, Dongdong and Ye, Yinyu},

如果您是本模型的贡献者，我们邀请您根据模型贡献文档，及时完善模型卡片内容。