This dataset includes 988 QandA extracted from Amazon Bedrock Documentation. It is then processed to match llama3-8B-Instruct template format.
It can be used to fine-tune llama3 not hallucinating about Amazon Bedrock. Let's see what Llama3-8B says about Amazon Bedrock!
''' base_model = "meta-llama/Meta-Llama-3-8B-Instruct" tokenizer = AutoTokenizer.from_pretrained(base_model) pipe = pipeline(task="text-generation"… See the full description on the dataset page:
https://huggingface.co/datasets/SepKeyPro/amazon-bedrock-ug-llama3-8B-Instruct-1k.
