File size: 4,552 Bytes
5dfa501
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f169ee9
5dfa501
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f169ee9
5dfa501
 
 
 
 
f169ee9
5dfa501
f169ee9
5dfa501
f169ee9
5dfa501
f169ee9
5dfa501
f169ee9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5dfa501
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: other
base_model: NousResearch/Meta-Llama-3-8B
tags:
- generated_from_trainer
model-index:
- name: out-llama8b-alpaca-data-pt-br
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
<details><summary>See axolotl config</summary>

axolotl version: `0.4.0`
```yaml
base_model: NousResearch/Meta-Llama-3-8B
model_type: LlamaForCausalLM
tokenizer_type: AutoTokenizer

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: dominguesm/alpaca-data-pt-br
    type: alpaca
dataset_prepared_path: last_run_prepared
val_set_size: 0.05
output_dir: ./out-llama8b-alpaca-data-pt-br

sequence_len: 8192
sample_packing: true
pad_to_sequence_len: true

wandb_project: meta-llama-8b-alpacadata-br
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 8
micro_batch_size: 1
num_epochs: 2
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 2e-5

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
early_stopping_patience:
resume_from_checkpoint:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 100
evals_per_epoch: 2
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
  pad_token: <|end_of_text|>

```

</details><br>

# out-llama8b-alpaca-data-pt-br

This model is a fine-tuned version of [NousResearch/Meta-Llama-3-8B](https://huggingface.co/NousResearch/Meta-Llama-3-8B) on the [dominguesm/alpaca-data-pt-br](https://huggingface.co/dominguesm/alpaca-data-pt-br) dataset.
It achieves the following results on the evaluation set:
- Loss: 1.1227

## Model description

The model is a Portuguese language understanding model designed to generate responses to a wide range of questions and prompts. It takes as input a natural language question or prompt and outputs a corresponding response.

The model is trained on a dataset of 51k examples, which is a cleaned and translated version of the original Alpaca Dataset released by Stanford. The original dataset was translated to Portuguese (Brazil) to provide a more culturally and linguistically relevant resource for the Brazilian market.

The dataset was carefully reviewed to identify and fix issues present in the original release, ensuring that the model is trained on high-quality data. The model is intended to be used in applications where a deep understanding of Portuguese language is required, such as chatbots, virtual assistants, and language translation systems.

## Intended uses:

Generating responses to natural language questions and prompts in Portuguese

Supporting chatbots, virtual assistants, and other conversational AI applications

Enhancing language translation systems and machine translation models

Providing a culturally and linguistically relevant resource for the Brazilian market

## Limitations

The model may not generalize well to other languages or dialects

The model may not perform well on out-of-domain or unseen topics

The model may not be able to handle ambiguous or open-ended prompts

The model may not be able to understand nuances of regional dialects or slang

The model may not be able to handle prompts that require common sense or real-world knowledge

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 100
- num_epochs: 2

### Training results

| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 1.382         | 0.01  | 1    | 1.4056          |
| 1.1762        | 0.5   | 45   | 1.1987          |
| 1.1294        | 0.99  | 90   | 1.1493          |
| 1.0028        | 1.47  | 135  | 1.1331          |
| 0.9899        | 1.97  | 180  | 1.1227          |


### Framework versions

- Transformers 4.40.0.dev0
- Pytorch 2.2.2+cu121
- Datasets 2.15.0
- Tokenizers 0.15.0