Habana
File size: 2,447 Bytes
95e6ca0
 
 
12c5034
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10d236f
 
9332301
10d236f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
12c5034
10d236f
 
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
---
license: apache-2.0
---

[Optimum Habana](https://github.com/huggingface/optimum-habana) is the interface between the Transformers library and Habana's Gaudi processor (HPU). It provides a set of tools enabling easy and fast model loading and fine-tuning on single- and multi-HPU settings for different downstream tasks.
Learn more about how to take advantage of the power of Habana HPUs to train Transformers models at [hf.co/hardware/habana](https://huggingface.co/hardware/habana).

## T5 model HPU configuration

This model contains just the `GaudiConfig` file for running the [T5](https://huggingface.co/t5-base) model on Habana's Gaudi processors (HPU).

**This model contains no model weights, only a GaudiConfig.**

This enables to specify:
- `use_habana_mixed_precision`: whether to use Habana Mixed Precision (HMP)
    - `hmp_opt_level`: optimization level for HMP, see [here](https://docs.habana.ai/en/latest/PyTorch/PyTorch_User_Guide/PT_Mixed_Precision.html#configuration-options) for a detailed explanation
    - `hmp_bf16_ops`: list of operators that should run in bf16
    - `hmp_fp32_ops`: list of operators that should run in fp32
    - `hmp_is_verbose`: verbosity
- `use_fused_adam`: whether to use Habana's custom AdamW implementation
- `use_fused_clip_norm`: whether to use Habana's fused gradient norm clipping operator

## Usage

The model is instantiated the same way as in the Transformers library.
The only difference is that there are a few new training arguments specific to HPUs.

[Here](https://github.com/huggingface/optimum-habana/blob/main/examples/summarization/run_summarization.py) is a summarization example script to fine-tune a model. You can run it with T5-small with the following command:
```bash
python run_summarization.py \
    --model_name_or_path t5-small \
    --do_train \
    --do_eval \
    --dataset_name cnn_dailymail \
    --dataset_config "3.0.0" \
    --source_prefix "summarize: " \
    --output_dir /tmp/tst-summarization \
    --per_device_train_batch_size 4 \
    --per_device_eval_batch_size 4 \
    --overwrite_output_dir \
    --predict_with_generate \
    --use_habana \
    --use_lazy_mode \
    --gaudi_config_name Habana/t5 \
    --ignore_pad_token_for_loss False \
    --pad_to_max_length \
    --save_strategy epoch \
    --throughput_warmup_steps 2
```

Check the [documentation](https://huggingface.co/docs/optimum/habana/index) out for more advanced usage and examples.