File size: 5,736 Bytes
db85f04
2ae2d8f
 
db85f04
2ae2d8f
 
 
db85f04
2ae2d8f
db85f04
2ae2d8f
 
 
 
 
 
db85f04
 
4f383c2
 
64e0234
045daa8
41eda79
db85f04
a3bda69
 
2ae2d8f
db85f04
2ae2d8f
db85f04
2ae2d8f
 
 
 
 
db85f04
2ae2d8f
db85f04
2ae2d8f
db85f04
2ae2d8f
db85f04
83b6426
 
 
 
 
 
 
 
 
2ae2d8f
 
 
 
 
 
 
 
 
 
41eda79
2ae2d8f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f4d7255
2ae2d8f
 
 
 
 
 
 
 
 
 
 
 
 
 
f4d7255
2ae2d8f
f4d7255
2ae2d8f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
db85f04
 
 
 
2ae2d8f
 
 
 
 
db85f04
 
 
 
 
 
 
 
 
 
 
 
2ae2d8f
 
 
db85f04
2ae2d8f
db85f04
2ae2d8f
db85f04
2ae2d8f
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
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
---
license: llama3
base_model: catallama/CataLlama-v0.1-Instruct-SFT
tags:
- llama
- llama-3
- Catalan
model-index:
- name: catallama/CataLlama-v0.1-Instruct-DPO
  results: []
datasets:
- catallama/Catalan-DPO
language:
- ca
- en
pipeline_tag: text-generation
---

![](https://huggingface.co/catallama/CataLlama-v0.1-Instruct-DPO/resolve/main/CataLlama-v0.1.png)

# CataLlama-v0.1-Instruct-DPO

**CataLlama-v0.1-Instruct-DPO** is a DPO fine-tune of [catallama/CataLlama-v0.1-Instruct-SFT](https://huggingface.co/catallama/CataLlama-v0.1-Instruct-SFT) on the [catallama/Catalan-DPO](https://huggingface.co/datasets/catallama/Catalan-DPO) dataset.

CataLlama was trained on roughly **445 million new tokens** in three separate stages. This is the 3rd and final stage of the training.

The model shows improved proficiency with the Catalan language.

**This is an instruction fine-tuned model, optimised with DPO, proficient on the following tasks in Catalan**

- *Information extraction (suitable for RAG)*
- *Named Entity Recognition (NER)*
- *Translation from English to Catalan and Catalan to English*
- *Summarization - both short form and long form*
- *Sentiment analysis*

**Model developers** [Laurentiu Petrea](https://www.linkedin.com/in/laurentiupetrea/) based on Llama-3 from Meta.

**Model Architecture** CataLlama is an auto-regressive language model that uses an optimized transformer architecture. The tuned versions use supervised fine-tuning (SFT) and direct preference optimisation (DPO) to align with human preferences for helpfulness and safety.

**License** The model uses the llama-3 license available at: [https://llama.meta.com/llama3/license](https://llama.meta.com/llama3/license)


## Benchmarks

| Benchmark          | Value  |
| ------------------ | ------ |
| MMLU 5 shot        | 47.34  |
| GSM8K cot 8 shot   | 43.29  |


### Use with transformers

See the snippet below for usage with Transformers:

**The model follows the same prompt template as Llama-3 Instruct**

```python
import transformers
import torch

model_id = "catallama/CataLlama-v0.1-Instruct-DPO"

pipeline = transformers.pipeline(
    "text-generation",
    model=model_id,
    model_kwargs={"torch_dtype": torch.bfloat16},
    device_map="auto",
)

messages = [
    {"role": "user", "content": "Ei com estàs avui?"},
]

prompt = pipeline.tokenizer.apply_chat_template(
    messages, 
    tokenize=False, 
    add_generation_prompt=True
)

outputs = pipeline(
    prompt,
    max_new_tokens=1024,
    do_sample=True,
    temperature=0.6,
    top_p=0.9,
)

print(outputs[0]["generated_text"][len(prompt):])
```

**Actual Inference Examples**

```
Prompt: Pot Donald Trump sopar amb Juli Cèsar?

Answer: No, Donald Trump no podria sopar amb Juli Cèsar, ja que Juli Cèsar va morir fa més de 2.000 anys i, per tant, no està viu per a poder sopar amb ell. A més, Juli Cèsar va ser un personatge històric i polític de l'antiga Roma, mentre que Donald Trump és el 45è president dels Estats Units, personatges que pertanyen a dues èpoques i cultures diferents. Per tant, la idea de que poguessin sopar junts és absurda i no té cap sentit històric ni lògic.
```

```
Prompt: A quina velocitat poden volar els cocodrils?

Answer: Els cocodrils no poden volar, ja que no tenen ales ni cap estructura que els permeti elevar-se per l'aire. Aquestes criatures aquàtiques depenen de la seva força i agilitat per moure's i caçar preses a terra ferma o al riu. No obstant això, poden assolir velocitats notables quan es tracta de fugir o perseguir una presa. El cocodril més ràpid, el cocodril negre africà, pot arribar a assolir velocitats de fins a 40 km/h en terra ferma, tot i que la seva velocitat a l'aigua és molt menor.
```

```
Prompt: Explica pas a pas com resoldre l'equació següent: 2x + 10 = 0

Answer:
Per resoldre l'equació 2x + 10 = 0, segueix els passos següents:

1. Aïllar la variable: per fer-ho, hem de restar 10 de tots dos costats de l'equació. Així, tenim:

2x = -10

2. Dividir tots dos costats per 2: això ens permet obtenir una equació en forma d'una sola incògnita. Així, tenim:

x = -5

3. Comprovar: sempre és recomanable comprovar la solució en l'equació original. Substituint x = -5 en 2x + 10 = 0, obtenim:

2(-5) + 10 = -10 + 10 = 0

Com que l'equació es compleix, la solució x = -5 és vàlida.
```


## Training procedure

The model was trained **with the same prompt template of Llama-3 Instruct**.

The model was trained for two epochs on **4x A100 80GB GPUs using DeepSpeed ZeRO** State-3 without CPU offloading.


### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 5e-07
- distributed_type: multi-GPU
- num_devices: 4
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 1


## Intended Use

**Note:** This model is not intended to beat benchmarks, but to demonstrate techniques for augmenting LLMs on new languages and preserve rare languages as part of our world heritage.

**Intended Use Cases** Llama 3 is intended for commercial and research use in English. Instruction tuned models are intended for assistant-like chat, whereas pretrained models can be adapted for a variety of natural language generation tasks.

**Out-of-scope** Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in any other way that is prohibited by the Acceptable Use Policy and Llama 3 Community License. Use in languages other than English**.

**Note: Developers may fine-tune Llama 3 models for languages beyond English provided they comply with the Llama 3 Community License and the Acceptable Use Policy.