provide the dummy data to train

[yezhengmao1]

provide the dummy data to train and test.
test the LoRA by the below code @Trilarflagz can test and modify the colab code to this? (maybe need better prompt and test instruction):

# ...
generation_config = model.generation_config
generation_config.max_new_tokens = 256
generation_config.temperature = 1
generation_config.top_p = 1
generation_config.num_return_sequences = 1
generation_config.pad_token_id = tokenizer.eos_token_id
generation_config.eos_token_id = tokenizer.eos_token_id

device = "cuda:0"
prompt = "Could you provide an introduction to m-LoRA?\n"
encoding = tokenizer(prompt, return_tensors="pt").to(device)
with torch.inference_mode():
  outputs = model.generate(
      input_ids = encoding.input_ids,
      attention_mask = encoding.attention_mask,
      generation_config = generation_config
  )

  full_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
  answer = full_text.replace(prompt, '').strip()
  print(answer)

[mikecovlee]

LGTM