Update README.md

README.md

@@ -1,3 +1,42 @@
----
-license: mit
----
+---
+license: mit
+datasets:
+- cnmoro/LexicalTriplets
+language:
+- en
+- pt
+pipeline_tag: feature-extraction
+library_name: sentence-transformers
+---
+
+This is a model trained on [cnmoro/LexicalTriplets](https://huggingface.co/datasets/cnmoro/LexicalTriplets) to produce lexical embeddings (not semantic!)
+
+This can be used to compute lexical similarity between words or phrases.
+
+Concept:
+"Some text" will be similar to "Sm txt"
+"King" will *not* be similar to "Queen" or "Royalty"
+"Dog" will *not* be similar to "Animal"
+"Doge" will be similar to "Dog"
+
+This will be trained for 2 epochs. The current model here is the first one.
+
+```python
+import torch
+from transformers import AutoModel, AutoTokenizer
+
+model_name = "cnmoro/LexicalEmbed-Base"
+
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModel.from_pretrained(model_name, trust_remote_code=True)
+model.eval()
+
+texts = ["hello world", "hel wor"]
+inputs = tokenizer(texts, padding=True, truncation=True, return_tensors="pt")
+
+with torch.no_grad():
+    embeddings = model(**inputs)
+
+cosine_sim = torch.nn.functional.cosine_similarity(embeddings[0], embeddings[1], dim=0)
+print(f"Cosine Similarity: {cosine_sim.item()}") # 0.8960
+```