Update README.md

README.md

@@ -1,3 +1,54 @@
----
-license: mit
----
+---
+license: apache-2.0
+tags:
+- pretrained
+- mistral
+- protein
+---
+
+# Model Card for Mistral-Prot-small (Mistral for protein)
+
+The Mistral-Prot-small Large Language Model (LLM) is a pretrained generative chemical molecule model with 16.725M parameters x 8 experts = 133.8M parameters. 
+It is derived from Mistral-7B-v0.1 model, which was simplified for protein: the number of layers and the hidden size were reduced. 
+The model was pretrained using 1M protein strings from the uniprot 50 database. 
+
+## Model Architecture
+
+Like Mistral-7B-v0.1, it is a transformer model, with the following architecture choices:
+- Grouped-Query Attention
+- Sliding-Window Attention
+- Byte-fallback BPE tokenizer
+
+## Load the model from huggingface:
+
+```
+import torch
+from transformers import AutoTokenizer, AutoModel
+
+tokenizer = AutoTokenizer.from_pretrained("RaphaelMourad/Mistral-Prot-small", trust_remote_code=True) 
+model = AutoModel.from_pretrained("RaphaelMourad/Mistral-Prot-small", trust_remote_code=True)
+```
+
+## Calculate the embedding of a DNA sequence
+
+```
+insulin = "MALWMRLLPLLALLALWGPDPAAAFVNQHLCGSHLVEALYLVCGERGFFYTPKTRREAEDLQVGQVELGGGPGAGSLQPLALEGSLQKRGIVEQCCTSICSLYQLENYCN"
+inputs = tokenizer(insulin, 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-Prot-small is a pretrained base model for chemistry.
+
+## Contact
+ 
+Raphaël Mourad. raphael.mourad@univ-tlse3.fr