Update README.md

README.md

@@ -45,6 +45,80 @@ language:
 
 # Salamandra Model Card
 
+## How to use
+
+> [!IMPORTANT]
+> This version of Salamandra is tailored exclusively for translation tasks. It lacks chat capabilities and has not been trained with any chat instructions.
+
+
+The instruction-following models use the commonly adopted ChatML template:
+
+```
+<|im_start|>system
+{SYSTEM PROMPT}<|im_end|>
+<|im_start|>user
+{USER PROMPT}<|im_end|>
+<|im_start|>assistant
+{MODEL RESPONSE}<|im_end|>
+<|im_start|>user
+[...]
+```
+
+The easiest way to apply it is by using the tokenizer's built-in functions, as shown in the following snippet.
+
+```python
+from datetime import datetime
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import transformers
+import torch
+
+model_id = "/gpfs/projects/bsc88/mt_translation/instructed_models/salamandraTA7b_instruct_mixture1/checkpoint-510"
+
+source = 'Spanish'
+target = 'Catalan'
+sentence = "Pensando en ti y en este amor que parte mi universo en dos y que llega del olvido hasta mi propia voz y araña mi pasado sin pedir perdón"
+ 
+text = f"Translate the following text from {source} into {target}.\n{source}: {sentence} \n{target}:"
+
+
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+
+stop_sequence = '<|im_end|>'
+eos_tokens = [tokenizer.eos_token_id,tokenizer.convert_tokens_to_ids(stop_sequence)]
+
+model = AutoModelForCausalLM.from_pretrained(
+    model_id,
+    device_map="auto",
+    torch_dtype=torch.bfloat16
+  )
+
+message = [ { "role": "user", "content": text } ]
+date_string = datetime.today().strftime('%Y-%m-%d')
+
+prompt = tokenizer.apply_chat_template(
+    message,
+    tokenize=False,
+    add_generation_prompt=True,
+    date_string=date_string
+)
+
+inputs = tokenizer.encode(prompt, add_special_tokens=False, return_tensors="pt")
+input_length = inputs.shape[1]
+outputs = model.generate(input_ids=inputs.to(model.device), 
+                         max_new_tokens=400,
+                         early_stopping=True,
+                         eos_token_id=eos_tokens, 
+                         pad_token_id=tokenizer.eos_token_id,
+                         num_beams=5)
+
+print(tokenizer.decode(outputs[0, input_length:], skip_special_tokens=True))
+# Pensant en tu i en aquest amor que parteix el meu univers en dos i que arriba des de l'oblit fins a la meva pròpia veu i esgarrapa el meu passat sense demanar perdó
+```
+
+
+Using this template, each turn is preceded by a `<|im_start|>` delimiter and the role of the entity 
+(either `user`, for content supplied by the user, or `assistant` for LLM responses), and finished with the `<|im_end|>` token.
+
 
 ## Data