vaaaas/chemcrow-fork

Chemcrow

ChemCrow is an open source package for the accurate solution of reasoning-intensive chemical tasks.

Built with Langchain, it uses a collection of chemical tools including RDKit, paper-qa, as well as some relevant databases in chemistry, like Pubchem and chem-space.

🤗 Try it out in HuggingFace!

⚠️ Note

This package does not contain all the tools described in the ChemCrow paper because of API usage restrictions. This repo will not give the same results as that paper.

All the experiments have been released under ChemCrow runs.

👩‍💻 Installation

pip install chemcrow

or for development:

conda create -n chemcrow python=3.9

conda activate chemcrow

CUDACXX=/usr/local/cuda-12.1/bin/nvcc CMAKE_ARGS="-DLLAMA_CUBLAS=on -DCMAKE_CUDA_ARCHITECTURES=native" FORCE_CMAKE=1 proxychains pip install llama-cpp-python --no-cache-dir --force-reinstall --upgrade

pip install -e .

pip install -r dev-requirements.txt

🔥 Usage

First set up your API keys in your environment.

export OPENAI_API_KEY=your-openai-api-key

You can optionally use Serp API:

export SERP_API_KEY=your-serpapi-api-key

In a Python session:

from chemcrow.agents import ChemCrow

chem_model = ChemCrow(model="gpt-4-0613", temp=0.1, streaming=False)
chem_model.run("What is the molecular weight of tylenol?")

💻 Running on a local machine

You can also use the program by loading a LlamaCpp (.gguf) or GPT4ALL (.bin) model as the LLM instead of using the OpenAI API.

from chemcrow.agents import ChemCrow

chem_model = ChemCrow(model="./models/llama-2-7b.Q8_0.gguf", 
                      tools_model="./models/llama-2-7b.Q8_0.gguf", 
                      temp=0.1, verbose=False, max_tokens=100, n_ctx=2048)
output = chem_model.run("What is the molecular weight of tylenol?")

>>> output
>>> The molecular weight of acetaminophen is 151.17 g/mol ...

✅ Citation

Bran, Andres M., et al. "ChemCrow: Augmenting large-language models with chemistry tools." arXiv preprint arXiv:2304.05376 (2023).

@article{bran2023chemcrow,
      title={ChemCrow: Augmenting large-language models with chemistry tools},
      author={Andres M Bran and Sam Cox and Oliver Schilter and Carlo Baldassari and Andrew D White and Philippe Schwaller},
      year={2023},
      eprint={2304.05376},
      archivePrefix={arXiv},
      primaryClass={physics.chem-ph},
      publisher={arXiv}
}