Update README.md

README.md

@@ -41,68 +41,62 @@ It achieves the following results on the evaluation set:
 ```
 conda install pytorch torchvision torchaudio pytorch-cuda=11.6 -c pytorch -c nvidia
 pip install optimum[openvino,nncf]==1.7.0
-cd examples/openvino/text-classification/
-pip install -r requirements.txt
+pip install datasets sentencepiece scipy scikit-learn protobuf evaluate
 pip install wandb # optional
 ```
 
-## NNCF config
+## Training script
 
-See `nncf_config.json` in this repo.
+See https://gist.github.com/yujiepan-work/5d7e513a47b353db89f6e1b512d7c080
 
 
 ## Run
 
 We use one card for training.
 
-```
-NNCFCFG=/path/to/nncf/config
+```bash
+NNCFCFG=/path/to/nncf_config/json
 python run_glue.py \
---lr_scheduler_type cosine_with_restarts \
---cosine_cycle_ratios 11,6 \
---cosine_cycle_decays 1,1 \
---save_best_model_after_epoch -1 \
---save_best_model_after_sparsity 0.7999 \
---model_name_or_path textattack/bert-base-uncased-SST-2 \
---teacher_model_or_path yoshitomo-matsubara/bert-large-uncased-sst2 \
---distillation_temperature 2 \
---task_name sst2 \
---nncf_compression_config $NNCFCFG \
---distillation_weight 0.95 \
---output_dir /tmp/bert-base-uncased-sst2-int8-unstructured80-17epoch \
---run_name bert-base-uncased-sst2-int8-unstructured80-17epoch \
---overwrite_output_dir \
---do_train \
---do_eval \
---max_seq_length 128 \
---per_device_train_batch_size 32 \
---per_device_eval_batch_size 32 \
---learning_rate 5e-05 \
---optim adamw_torch \
---num_train_epochs 17 \
---logging_steps 1 \
---evaluation_strategy steps \
---eval_steps 250 \
---save_strategy steps \
---save_steps 250 \
---save_total_limit 1 \
---fp16 \
---seed 1
+  --lr_scheduler_type cosine_with_restarts \
+  --cosine_lr_scheduler_cycles 11 6 \
+  --record_best_model_after_epoch 9 \
+  --load_best_model_at_end True \
+  --metric_for_best_model accuracy \
+  --model_name_or_path textattack/bert-base-uncased-SST-2 \
+  --teacher_model_or_path yoshitomo-matsubara/bert-large-uncased-sst2 \
+  --distillation_temperature 2 \
+  --task_name sst2 \
+  --nncf_compression_config $NNCFCFG \
+  --distillation_weight 0.95 \
+  --output_dir /tmp/bert-base-uncased-sst2-int8-unstructured80 \
+  --overwrite_output_dir \
+  --run_name bert-base-uncased-sst2-int8-unstructured80 \
+  --do_train \
+  --do_eval \
+  --max_seq_length 128 \
+  --per_device_train_batch_size 32 \
+  --per_device_eval_batch_size 32 \
+  --learning_rate 5e-05 \
+  --optim adamw_torch \
+  --num_train_epochs 17 \
+  --logging_steps 1 \
+  --evaluation_strategy steps \
+  --eval_steps 250 \
+  --save_strategy steps \
+  --save_steps 250 \
+  --save_total_limit 1 \
+  --fp16 \
+  --seed 1
 ```
 
-The best model checkpoint is stored in the `best_model` folder. Here we only upload that checkpoint folder together with some config files.
-
-
-## inference
-
-https://gist.github.com/yujiepan-work/c38dc4e56c7a9d803c42988f7b7d260a
-
-
 ### Framework versions
 
 - Transformers 4.26.0
 - Pytorch 1.13.1+cu116
 - Datasets 2.8.0
 - Tokenizers 0.13.2
+- Optimum 1.6.3
+- Optimum-intel 1.7.0
+- NNCF 2.4.0
 
 For a full description of the environment, please refer to `pip-requirements.txt` and `conda-requirements.txt`.