ur-whitelab / chemcrow-public Goto Github PK

View Code? Open in Web Editor NEW

519.0 519.0 71.0 5.91 MB

Chemcrow

License: MIT License

Python 100.00%

chemcrow-public's People

Contributors

Stargazers

Watchers

Forkers

teletchealab kimist99 ardeat eperrier lilleswing vinayakhegde666 waseemabbas05 stjordanis star7js jeffhsu3 mkmicikiewiczselvita yangqiantian luca-git standardgalactic gaybro8777 jameshwade dingchaoz llmr-boringtao cormac-c kp-forks datastark npokemon ahmedelmaamounamin aehsan4004 anuragatulya starleejose cklam12345 avanteijlingen tbabcock22 tayyabmujahid cephren juliev42 raven-biosciences mikegongolidis ftgreat praljakreps sivadasetty allthingsllm apollohuang1 chimiadao pranavjaintridiagonal ccc0168 sorokinvld crazyboystop schwallergroup datnoor keceli yskale huangyingting assistaplus xc0r brunoscaglione ritesh001 liaojf24 adrianpuiu gareththomasnz nhattruongpham krish240574 dvanrom1 catalyst-plus crjaensch aiseei natural20ai jensssen roy029 raoulbia-ai qilongxu michaelcshn

chemcrow-public's Issues

Add remaining tools

We should add in all remaining tools including safety, rxn_predict, etc.

Paperscraper not in requirements. LitSearch tool fails

paperscraper is not in requirements, so your sample code fails with:

[/usr/local/lib/python3.10/dist-packages/chemcrow/tools/search.py](https://localhost:8080/#) in <module>
      3 import paperqa
      4 import langchain
----> 5 import paperscraper
      6 from langchain.tools import BaseTool
      7 from langchain import SerpAPIWrapper, OpenAI

ModuleNotFoundError: No module named 'paperscraper'

When installing the latest version, I get:
AttributeError: module 'paperscraper' has no attribute 'search_papers'

What version of paperscraper are you using?

Add paperqa :)

langchain 0.1.x support

Hi. I was wondering if there would be support of langchain 0.1.x versions. I'm trying to launch chemcrow with other langchain agents. Thank you

evaluate code

I read your experimental results in detail, but I didn't find the evaluation part of the code in the code. Can you share it?

RXNRetrosynthesis failing attempts

Hi, thanks for a truly exciting repository.

I have obtained an RXN4CHEM_API_KEY and I am trying to run chemcrow to give suggest a synthesis plan for a given molecule.

However, code fails after multiple attempts:

ReactionRetrosynthesis
Action Input:
CCOC(=O)c1cnc2c(C)cc(C)cc2c1Nc1ccc(OC)c(OC)c1Exception thrown when attempting to run <function RXNRetrosynthesis.predict_retrosynthesis at 0x124731ea0>, attempt 0 of 10
Exception thrown when attempting to run <function RXNRetrosynthesis.predict_retrosynthesis at 0x124731ea0>, attempt 1 of 10
Exception thrown when attempting to run <function RXNRetrosynthesis.predict_retrosynthesis at 0x124731ea0>, attempt 2 of 10
Exception thrown when attempting to run <function RXNRetrosynthesis.predict_retrosynthesis at 0x124731ea0>, attempt 3 of 10
Exception thrown when attempting to run <function RXNRetrosynthesis.predict_retrosynthesis at 0x124731ea0>, attempt 4 of 10
Exception thrown when attempting to run <function RXNRetrosynthesis.predict_retrosynthesis at 0x124731ea0>, attempt 5 of 10
Exception thrown when attempting to run <function RXNRetrosynthesis.predict_retrosynthesis at 0x124731ea0>, attempt 6 of 10
Exception thrown when attempting to run <function RXNRetrosynthesis.predict_retrosynthesis at 0x124731ea0>, attempt 7 of 10
Exception thrown when attempting to run <function RXNRetrosynthesis.predict_retrosynthesis at 0x124731ea0>, attempt 8 of 10
Exception thrown when attempting to run <function RXNRetrosynthesis.predict_retrosynthesis at 0x124731ea0>, attempt 9 of 10
Traceback (most recent call last):
 File "/Users/mduranfrigola/Documents/GitHub/ai-intro-workshop/chemcrowdev.py", line 23, in <module>
  result = mrkl.run(prompt)
 File "/Users/mduranfrigola/miniconda3/envs/chemcrow/lib/python3.10/site-packages/chemcrow/agents/chemcrow.py", line 84, in run
  outputs = self.agent_executor({"input": prompt})
 File "/Users/mduranfrigola/miniconda3/envs/chemcrow/lib/python3.10/site-packages/langchain/chains/base.py", line 282, in __call__
  raise e
 File "/Users/mduranfrigola/miniconda3/envs/chemcrow/lib/python3.10/site-packages/langchain/chains/base.py", line 276, in __call__
  self._call(inputs, run_manager=run_manager)
 File "/Users/mduranfrigola/miniconda3/envs/chemcrow/lib/python3.10/site-packages/langchain/agents/agent.py", line 1036, in _call
  next_step_output = self._take_next_step(
 File "/Users/mduranfrigola/miniconda3/envs/chemcrow/lib/python3.10/site-packages/langchain/agents/agent.py", line 891, in _take_next_step
  observation = tool.run(
 File "/Users/mduranfrigola/miniconda3/envs/chemcrow/lib/python3.10/site-packages/langchain/tools/base.py", line 351, in run
  raise e
 File "/Users/mduranfrigola/miniconda3/envs/chemcrow/lib/python3.10/site-packages/langchain/tools/base.py", line 325, in run
  else self._run(*tool_args, **tool_kwargs)
 File "/Users/mduranfrigola/miniconda3/envs/chemcrow/lib/python3.10/site-packages/chemcrow/tools/rxn4chem.py", line 147, in _run
  prediction_id = self.predict_retrosynthesis(target)
 File "/Users/mduranfrigola/miniconda3/envs/chemcrow/lib/python3.10/site-packages/chemcrow/tools/rxn4chem.py", line 75, in newfn
  return func(*args, **kwargs)
 File "/Users/mduranfrigola/miniconda3/envs/chemcrow/lib/python3.10/site-packages/chemcrow/tools/rxn4chem.py", line 170, in predict_retrosynthesis
  raise KeyError
KeyError

Do you know what may be going on?

Thanks so much in advance.

ERROR: Failed building wheel for molbloom

Got this error while simply installing using PIP

Add similarity compare in safety tool

is self.rephrase_chain used for anything?

the declaration in chemcrow.py

    self.rephrase_chain = chains.LLMChain(prompt=rephrase, llm=self.llm)

don't seem to be ever used for anything. Why is it there?

It actually would be nice to see it used in some fashion to build up a retrying chain.

Wondering if exists token exceed problem

Thanks for ur excellent work, very insightful. But I'm wondering: with 13 tools and local knowledge sources, it seems very easy to reach the 16k token limit(turbo-0613)，how to avoid these problems?

Not working with langchain due to changes and migration to langchain experimental

As of today I am using langchain perfectly with other projects. Just testing chemprop minimum script

from chemcrow.agents import ChemCrow

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

Get the error

/home/stephany/.local/lib/python3.9/site-packages/langchain/__init__.py:34: UserWarning: Importing SerpAPIWrapper from langchain root module is no longer supported. Please use langchain.utilities.SerpAPIWrapper instead.
  warnings.warn(
/home/stephany/.local/lib/python3.9/site-packages/langchain/__init__.py:34: UserWarning: Importing LLMChain from langchain root module is no longer supported. Please use langchain.chains.LLMChain instead.
  warnings.warn(
/home/stephany/.local/lib/python3.9/site-packages/langchain/__init__.py:34: UserWarning: Importing PromptTemplate from langchain root module is no longer supported. Please use langchain.prompts.PromptTemplate instead.
  warnings.warn(
Traceback (most recent call last):
  File "/home/stephany/.local/lib/python3.9/site-packages/chemcrow/agents/chemcrow.py", line 46, in __init__
    self.llm = _make_llm(model, temp, verbose, openai_api_key)
  File "/home/stephany/.local/lib/python3.9/site-packages/chemcrow/agents/chemcrow.py", line 12, in _make_llm
    llm = langchain.chat_models.ChatOpenAI(
  File "/home/stephany/.local/lib/python3.9/site-packages/langchain/load/serializable.py", line 97, in __init__
    super().__init__(**kwargs)
  File "/home/stephany/.local/lib/python3.9/site-packages/pydantic/v1/main.py", line 341, in __init__
    raise validation_error
pydantic.v1.error_wrappers.ValidationError: 1 validation error for ChatOpenAI
__root__
  Did not find openai_api_key, please add an environment variable `OPENAI_API_KEY` which contains it, or pass  `openai_api_key` as a named parameter. (type=value_error)

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/stephany/test_query/latest/testing.py", line 3, in <module>
    chem_model = ChemCrow(model="gpt-4-0613", temp=0.1, verbose=True)
  File "/home/stephany/.local/lib/python3.9/site-packages/chemcrow/agents/chemcrow.py", line 48, in __init__
    raise ValueError('Invalid OpenAI API key')
ValueError: Invalid OpenAI API key

ModuleNotFoundError: No module named 'streamlit.delta_generator'; 'streamlit' is not a package

I'm getting ModuleNotFoundError: No module named 'streamlit.delta_generator'; 'streamlit' is not a package
when trying to init ChemCrow

Not working due to: No module named 'langchain_core.schema'

Installation and API key setup was done as described in the readme, but when running

from chemcrow.agents import ChemCrow

I am getting the error

ModuleNotFoundError: No module named 'langchain_core.schema'

langchain is installed and version confirmed to be 0.0.275.

Thanks in advance!

Add more tests

We still need tests for the following tools:

```
    ExplosiveCheck,
```
```
    ControlChemCheck,
```
```
    SafetySummary,
```
```
    LitSearch
```

pip install chemcrow gives version 0.2.0 instead of 0.3.0

https://pypi.org/project/chemcrow/

WARNING:paperscraper.load_dumps: No dump found for biorxiv. Skipping entry.

Dear Sir

Upon executing the python test_agent.py command, I encountered the following warnings:

    WARNING:paperscraper.load_dumps: No dump found for biorxiv. Skipping entry.
    WARNING:paperscraper.load_dumps: No dump found for chemrxiv. Skipping entry.
    WARNING:paperscraper.load_dumps: No dump found for medrxiv. Skipping entry.
    WARNING:paperscraper.load_dumps: No dumps found for either biorxiv or medrxiv. Consider using paperscraper.get_dumps.* to fetch the dumps.

Could you please assist me in resolving this issue? It appears that the necessary data dumps are missing, and I am unsure how to proceed with using paperscraper.get_dumps.* to retrieve them.

Thanks!