|
--- |
|
license: apache-2.0 |
|
--- |
|
|
|
```` |
|
from transformers import AutoTokenizer, AutoModelForCausalLM |
|
|
|
tokenizer = AutoTokenizer.from_pretrained("RootYuan/opt-1.3b-alpaca") |
|
model = AutoModelForCausalLM.from_pretrained("RootYuan/opt-1.3b-alpaca") |
|
```` |
|
usage: |
|
```` |
|
instruction = "Classify the following into animals, plants, and minerals" |
|
input = "Oak tree, copper ore, elephant" |
|
|
|
prompts_no_input = f"### Instruction:\n{instruction}\n\n### Response:" |
|
prompts_with_input = f"### Instruction:\n{instruction}\n\n### Input:\n{input}\n\n### Response:" |
|
prompts = prompts_no_input if input is None else prompts_with_input |
|
|
|
inputs = tokenizer.encode(prompts, return_tensors="pt") |
|
outputs = model.generate(inputs, max_new_tokens=64) |
|
ans = tokenizer.decode(outputs[0]).strip('</s>')[len(prompts):] |
|
if input is None: |
|
print(f"Human: {instruction}") |
|
else: |
|
print(f"Human: {instruction}\nInput: {input}") |
|
print(f"Assistant: {ans}") |
|
```` |
|
outputs: |
|
```` |
|
Human: Classify the following into animals, plants, and minerals |
|
Input: Oak tree, copper ore, elephant |
|
Assistant: Oak tree: Plant |
|
Copper ore: Mineral |
|
Elephant: Animal |
|
```` |
|
|