|
--- |
|
license: apache-2.0 |
|
tags: |
|
- pretrained |
|
- mistral |
|
- chemistry |
|
--- |
|
|
|
# Model Card for Mistral-Chem-v1-15M (Mistral for chemistry) |
|
|
|
The Mistral-Chem-v1-15M Large Language Model (LLM) is a pretrained generative chemical molecule model with 1.9M parameters x 8 experts = 15.2M parameters. |
|
It is derived from Mixtral-8x7B-v0.1 model, which was simplified for molecules: the number of layers and the hidden size were reduced. |
|
The model was pretrained using 10M molecule SMILES strings from the ZINC 15 database. |
|
|
|
## Model Architecture |
|
|
|
Like Mixtral-8x7B-v0.1, it is a transformer model, with the following architecture choices: |
|
- Grouped-Query Attention |
|
- Sliding-Window Attention |
|
- Byte-fallback BPE tokenizer |
|
- Mixture of Experts |
|
|
|
## Load the model from huggingface: |
|
|
|
``` |
|
import torch |
|
from transformers import AutoTokenizer, AutoModel |
|
|
|
tokenizer = AutoTokenizer.from_pretrained("RaphaelMourad/Mistral-Chem-v1-15M", trust_remote_code=True) |
|
model = AutoModel.from_pretrained("RaphaelMourad/Mistral-Chem-v1-15M", trust_remote_code=True) |
|
``` |
|
|
|
## Calculate the embedding of a DNA sequence |
|
|
|
``` |
|
chem = "CCCCC[C@H](Br)CC" |
|
inputs = tokenizer(chem, return_tensors = 'pt')["input_ids"] |
|
hidden_states = model(inputs)[0] # [1, sequence_length, 256] |
|
|
|
# embedding with max pooling |
|
embedding_max = torch.max(hidden_states[0], dim=0)[0] |
|
print(embedding_max.shape) # expect to be 256 |
|
``` |
|
|
|
## Troubleshooting |
|
|
|
Ensure you are utilizing a stable version of Transformers, 4.34.0 or newer. |
|
|
|
## Notice |
|
|
|
Mistral-Chem-v1-15M is a pretrained base model for chemistry. |
|
|
|
## Contact |
|
|
|
Raphaël Mourad. [email protected] |
