File size: 2,549 Bytes
ed21a70
 
 
 
 
 
 
 
 
 
 
9425e9c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
---
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). We provide:

* 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


```