license: bigscience-bloom-rail-1.0
language:
- es
- en
pipeline_tag: text-generation
library_name: transformers
tags:
- alpaca
- bloom
- LLM
datasets:
- tatsu-lab/alpaca
inference: false
widget:
- text: >-
Below is an instruction that describes a task, paired with an input that
provides further context.
Write a response that appropriately completes the request.
### Instruction:
Tell me about alpacas
![](https://huggingface.co/platzi/chivoom/resolve/main/chivoom_logo-removebg-preview.png)
Chivoom: Spanish Alpaca (Chiva) 馃悙 + BLOOM 馃挳
IMPORTANT: This is just a PoC and still WIP!
Adapter Description
This adapter was created with the PEFT library and allowed the base model BigScience/BLOOM 7B1 to be fine-tuned on the Stanford's Alpaca Dataset (translated to Spanish) by using the method LoRA.
Model Description
BigScience Large Open-science Open-access Multilingual Language Model
Training data
We translated to Spanish the Alpaca dataset.
Alpaca is a dataset of 52,000 instructions and demonstrations generated by OpenAI's text-davinci-003
engine. This instruction data can be used to conduct instruction-tuning for language models and make the language model follow instruction better.
The authors built on the data generation pipeline from Self-Instruct framework and made the following modifications:
- The
text-davinci-003
engine to generate the instruction data instead ofdavinci
. - A new prompt was written that explicitly gave the requirement of instruction generation to
text-davinci-003
. - Much more aggressive batch decoding was used, i.e., generating 20 instructions at once, which significantly reduced the cost of data generation.
- The data generation pipeline was simplified by discarding the difference between classification and non-classification instructions.
- Only a single instance was generated for each instruction, instead of 2 to 3 instances as in Self-Instruct.
This produced an instruction-following dataset with 52K examples obtained at a much lower cost (less than $500). In a preliminary study, the authors also found that the 52K generated data to be much more diverse than the data released by Self-Instruct.
Training procedure
TBA
How to use
import torch
from peft import PeftModel, PeftConfig
from transformers import AutoModelForCausalLM, AutoTokenizer, GenerationConfig
peft_model_id = "platzi/chivoom"
config = PeftConfig.from_pretrained(peft_model_id)
model = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path, return_dict=True, load_in_8bit=True, device_map="auto")
tokenizer = AutoTokenizer.from_pretrained("bigscience/bloom-7b1")
model = PeftModel.from_pretrained(model, peft_model_id)
model.eval()
# Based on the inference code by `tloen/alpaca-lora`
def generate_prompt(instruction, input=None):
if input:
return f"""A continuaci贸n se muestra una instrucci贸n que describe una tarea, emparejada con una entrada que proporciona m谩s contexto. Escribe una respuesta que complete adecuadamente la petici贸n.
### Instrucci贸n:
{instruction}
### Entrada:
{input}
### Respuesta:"""
else:
return f"""A continuaci贸n se muestra una instrucci贸n que describe una tarea. Escribe una respuesta que complete adecuadamente la petici贸n.
### Instrucci贸n:
{instruction}
### Respuesta:"""
def generate(
instruction,
input=None,
temperature=0.1,
top_p=0.75,
top_k=40,
num_beams=4,
**kwargs,
):
prompt = generate_prompt(instruction, input)
inputs = tokenizer(prompt, return_tensors="pt")
input_ids = inputs["input_ids"].cuda()
generation_config = GenerationConfig(
temperature=temperature,
top_p=top_p,
top_k=top_k,
num_beams=num_beams,
**kwargs,
)
with torch.no_grad():
generation_output = model.generate(
input_ids=input_ids,
generation_config=generation_config,
return_dict_in_generate=True,
output_scores=True,
max_new_tokens=256,
)
s = generation_output.sequences[0]
output = tokenizer.decode(s)
return output.split("### Response:")[1]
instruction = "驴Qu茅 es un chivo?"
print("Instrucci贸n:", instruction)
print("Respuesta:", generate(instruction))