Update README.md

README.md

@@ -14,6 +14,8 @@ SEA-LION is a collection of Large Language Models (LLMs) which has been pretrain
 Llama3 8B CPT SEA-Lionv2.1 Instruct is a multilingual model which has been fine-tuned with around **100,000 English instruction-completion pairs** alongside a smaller pool of around **50,000 instruction-completion pairs** from other ASEAN languages, such as Indonesian, Thai and Vietnamese.
 These instructions have been carefully curated and rewritten to ensure the model was trained on truly open, commercially permissive and high quality datasets.
 
+Llama3 8B CPT SEA-Lionv2.1 Instruct has undergone additional supervised fine-tuning and alignment compared to the now deprecated Llama3 8B CPT SEA-Lionv2 Instruct. These improvements have increased the model's capabilities in chat interactions and its ability to follow instructions accurately.
+
 SEA-LION stands for _Southeast Asian Languages In One Network_.
 
 - **Developed by:** Products Pillar, AI Singapore
@@ -86,9 +88,32 @@ outputs = pipeline(
 print(outputs[0]["generated_text"][-1])
 ```
 
-### Prompting Guide
+### Accessing Older Revisions
+Huggingface provides support for the revision parameter, allowing users to access specific versions of models. This can be used to retrieve the original llama3-8b-cpt-SEA-Lionv2-instruct model with the tag "v2.0.0".
+```python
+# Please use transformers==4.43.2
+
+import transformers
+import torch
+
+model_id = "aisingapore/llama3-8b-cpt-SEA-Lionv2.1-instruct"
 
-_Coming soon_
+pipeline = transformers.pipeline(
+    "text-generation",
+    model=model_id,
+    model_kwargs={"torch_dtype": torch.bfloat16, "revision": "v2.0.0"},
+    device_map="auto",
+)
+messages = [
+    {"role": "user", "content": "Apa sentimen dari kalimat berikut ini?\nKalimat: Buku ini sangat membosankan.\nJawaban: "},
+]
+
+outputs = pipeline(
+    messages,
+    max_new_tokens=256,
+)
+print(outputs[0]["generated_text"][-1])
+```
 
 ### Caveats
 It is important for users to be aware that our model exhibits certain limitations that warrant consideration. Like many LLMs, the model can hallucinate and occasionally generates irrelevant content, introducing fictional elements that are not grounded in the provided context. Users should also exercise caution in interpreting and validating the model's responses due to the potential inconsistencies in its reasoning.