Downloads · 30 days
50
2% of all-time downloads
RichardErkhov/yerevann_-_chemma-2b-gguf
yerevann_-_chemma-2b-gguf is a machine learning model from RichardErkhov. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Downloads · 30 days
50
2% of all-time downloads
All-time downloads
2.3K
Public
Repo size
30.7 GB
Likes
0
Public
Click a slice to open those files.
.gguf35.6 GB · 100%
From the Hugging Face model README
Quantization made by Richard Erkhov.
chemma-2b - GGUF
| Name | Quant method | Size |
|---|---|---|
| chemma-2b.Q2_K.gguf | Q2_K | 1.08GB |
| chemma-2b.IQ3_XS.gguf | IQ3_XS | 1.16GB |
| chemma-2b.IQ3_S.gguf | IQ3_S | 1.2GB |
| chemma-2b.Q3_K_S.gguf | Q3_K_S | 1.2GB |
| chemma-2b.IQ3_M.gguf | IQ3_M | 1.22GB |
| chemma-2b.Q3_K.gguf | Q3_K | 1.29GB |
| chemma-2b.Q3_K_M.gguf | Q3_K_M | 1.29GB |
| chemma-2b.Q3_K_L.gguf | Q3_K_L | 1.36GB |
| chemma-2b.IQ4_XS.gguf | IQ4_XS | 1.4GB |
| chemma-2b.Q4_0.gguf | Q4_0 | 1.44GB |
| chemma-2b.IQ4_NL.gguf | IQ4_NL | 1.45GB |
| chemma-2b.Q4_K_S.gguf | Q4_K_S | 1.45GB |
| chemma-2b.Q4_K.gguf | Q4_K | 1.52GB |
| chemma-2b.Q4_K_M.gguf | Q4_K_M | 1.52GB |
| chemma-2b.Q4_1.gguf | Q4_1 | 1.56GB |
| chemma-2b.Q5_0.gguf | Q5_0 | 1.68GB |
| chemma-2b.Q5_K_S.gguf | Q5_K_S | 1.68GB |
| chemma-2b.Q5_K.gguf | Q5_K | 1.71GB |
| chemma-2b.Q5_K_M.gguf | Q5_K_M | 1.71GB |
| chemma-2b.Q5_1.gguf | Q5_1 | 1.79GB |
| chemma-2b.Q6_K.gguf | Q6_K | 1.92GB |
| chemma-2b.Q8_0.gguf | Q8_0 | 2.49GB |
license: cc-by-nc-4.0 language:
Chemma-2B is a continually pretrained gemma-2b model for organic molecules. It is pretrained on (soon-to-be-released) 40B tokens covering 110M+ molecules from PubChem as well as their chemical properties (molecular weight, synthetic accessibility score, drug-likeness etc.) and similarities (Tanimoto distance between ECFP fingerprints).
Example prompts:
</s>[START_SMILES]CC(=O)OC1=CC=CC=C1C(=O)O[END_SMILES][SAS] will attempt to predict the synthetic accessibility score of the given molecule.
</s>[SAS]2.25[/SAS][SIMILAR]0.62 CC(=O)OC1=CC=CC=C1C(=O)O[/SIMILAR][START_SMILES] will attempt to generate a molecule that has 2.25 SAS score and
has a 0.62 similarity score to the given molecule.
The model can be wrapped into an optimization loop to traverse the chemical space with evolving prompts.
A preprint with the details of the model and an optimization algorithm built on top of this model that sets state-of-the-art on Practical Molecular Optimization and other benchmarks will be released soon.
Few notes:
</s> symbol.rdkit.[CLOGP], [WEIGHT], [QED], [SAS], [TPSA], [RINGCOUNT], [SIMILAR]...The model is part of the 3-model family: Chemlactica-125M, Chemlactica-1.3B and Chemma-2B.
We are looking forward to see the community using the model in new applications and contexts.