README.md

metadata

license: apache-2.0

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.

T5 model HPU configuration

This model contains just the GaudiConfig file for running the T5 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 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 is a summarization example script to fine-tune a model. You can run it with T5-small with the following command:

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 out for more advanced usage and examples.