|
--- |
|
datasets: |
|
- ysharma/short_jokes |
|
language: |
|
- en |
|
metrics: |
|
- accuracy |
|
library_name: peft |
|
pipeline_tag: text-generation |
|
tags: |
|
- joke |
|
--- |
|
|
|
#### Fine-tuning examples |
|
|
|
You can find fine-tuning notebooks under the [`examples/` directory](https://huggingface.co/google/gemma-7b/tree/main/examples). |
|
|
|
* A notebook that you can run on a free-tier Google Colab instance to perform SFT on English quotes dataset |
|
|
|
|
|
#### Running the model on a CPU |
|
|
|
|
|
```python |
|
from transformers import AutoTokenizer, AutoModelForCausalLM |
|
|
|
tokenizer = AutoTokenizer.from_pretrained("google/gemma-7b") |
|
model = AutoModelForCausalLM.from_pretrained("google/gemma-7b") |
|
|
|
input_text = "Write me a poem about Machine Learning." |
|
input_ids = tokenizer(input_text, return_tensors="pt") |
|
|
|
outputs = model.generate(**input_ids) |
|
print(tokenizer.decode(outputs[0])) |
|
``` |
|
|
|
#### Running the model on a GPU After Finetune_model |
|
|
|
```python |
|
import torch |
|
from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig |
|
from peft import PeftModel |
|
|
|
base_model_id= "google/gemma-2b" |
|
bnb_config = BitsAndBytesConfig( |
|
load_in_4bit=True, |
|
bnb_4bit_use_double_quant=True, |
|
bnb_4bit_quant_type="nf4", |
|
bnb_4bit_compute_dtype=torch.bfloat16 |
|
) |
|
|
|
base_model = AutoModelForCausalLM.from_pretrained( |
|
base_model_id, # Mistral, same as before |
|
quantization_config=bnb_config, # Same quantization config as before |
|
device_map="auto", |
|
trust_remote_code=True, |
|
) |
|
|
|
eval_tokenizer = AutoTokenizer.from_pretrained(base_model_id, add_bos_token=True, trust_remote_code=True) |
|
|
|
ft_model = PeftModel.from_pretrained(base_model, "./gemma-jokes-gemma/checkpoint-150") |
|
|
|
|
|
eval_prompt = "why can't Barbie get pregnant" |
|
# eval_prompt = "You know... When someone says to you Jesus loves you It's always comforting. Unless you are in a Mexican jail." |
|
model_input = eval_tokenizer(eval_prompt, return_tensors="pt").to("cuda:0") |
|
ft_model.eval() |
|
with torch.no_grad(): |
|
print(eval_tokenizer.decode(ft_model.generate(**model_input, max_new_tokens=100, repetition_penalty=1.15)[0], skip_special_tokens=True)) |
|
|
|
# Result |
|
# why can't Barbie get pregnant? Because she has no eggs. |
|
|
|
# Why did the chicken cross the road? To get to the other side of the egg. |
|
|
|
# Why do chickens lay eggs in their sleep? Because they don't want to wake up and find out they're dead. |
|
|
|
# Why do chickens wear glasses? Because they have a hard time seeing the yolk. |
|
|
|
# Why do chickens eat so much? Because they are always hungry. |
|
|
|
# Why do chickens like to go to the beach? Because they love laying eggs |
|
|
|
|
|
``` |