diff --git a/convlab2/base_models/t5/key2gen/create_data_key2gen.py b/convlab2/base_models/t5/key2gen/create_data_key2gen.py
new file mode 100644
index 0000000000000000000000000000000000000000..edf4a130997f9d6ed0716f1e99577fb4621e8cb5
--- /dev/null
+++ b/convlab2/base_models/t5/key2gen/create_data_key2gen.py
@@ -0,0 +1,131 @@
+import os
+import json
+from tqdm import tqdm
+import re
+from transformers import AutoTokenizer
+from convlab2.util import load_dataset, load_nlu_data, load_dst_data, load_policy_data, load_nlg_data, load_e2e_data, load_rg_data
+from convlab2.base_models.t5.nlu.serialization import serialize_dialogue_acts, deserialize_dialogue_acts, equal_da_seq
+from collections import Counter
+
+def create_nlg_data(dataset, data_dir, args):
+ data_by_split = load_nlu_data(dataset, speaker=args.speaker, use_context=args.context_window_size>0, context_window_size=args.context_window_size)
+ data_dir = os.path.join(data_dir, args.speaker, f'context_{args.context_window_size}')
+ os.makedirs(data_dir, exist_ok=True)
+
+ data_splits = data_by_split.keys()
+ for data_split in data_splits:
+ data = []
+ for sample in tqdm(data_by_split[data_split], desc=f'{data_split} sample', leave=False):
+ if args.key2gen:
+ da_dict = {}
+ for da_type in sample['dialogue_acts']:
+ for da in sample['dialogue_acts'][da_type]:
+ intent, domain, slot, value = da['intent'], da['domain'], da['slot'], da.get('value', '')
+ intent_domain = f'{intent}-{domain}'
+ da_dict.setdefault(intent_domain, [])
+ da_dict[intent_domain].append((slot, value))
+ keywords = []
+ for intent_domain, slot_values in da_dict.items():
+ keywords.append(intent_domain)
+ for slot, value in slot_values:
+ if len(slot) > 0:
+ keywords.append(slot)
+ if len(value) > 0:
+ keywords.append(value)
+ dialogue_acts_seq = ' | '.join(keywords)
+ else:
+ dialogue_acts_seq = serialize_dialogue_acts(sample['dialogue_acts'])
+
+ if args.context_window_size>0:
+ context = '\n'.join([f"{turn['speaker']}: {turn['utterance']}" for turn in sample['context']]+[f'{sample["speaker"]}: '])
+ context = f'{dialogue_acts_seq}\n\ncontext: {context}'
+ else:
+ context = f'{dialogue_acts_seq}\n\ncontext: {sample["speaker"]}: '
+ data.append(json.dumps({'context+da': context, 'response': sample['utterance']}, ensure_ascii=False)+'\n')
+
+ file_name = os.path.join(data_dir, f"{data_split}.json")
+ with open(file_name, "w", encoding='utf-8') as f:
+ f.writelines(data)
+ data_by_split[data_split] = data
+ return data_by_split
+
+def create_dart_data(dataset, data_dir, args):
+ data_by_split = dataset
+ os.makedirs(data_dir, exist_ok=True)
+
+ data_splits = data_by_split.keys()
+ for data_split in data_splits:
+ data = []
+ for sample in tqdm(data_by_split[data_split], desc=f'{data_split} sample', leave=False):
+ sample = sample['turns'][0]
+ triples = sample['tripleset']
+ if args.key2gen:
+ keywords = [w for triple in triples for w in triple]
+ # TODO: try adding prompt
+ # entity_cnt = Counter()
+ # for triple in triples:
+ # e1, r, e2 = triple
+ # for e in [e1, e2]:
+ # if e.startswith('[') and e.endswith(']'):
+ # continue
+ # entity_cnt[e] += 1
+
+ # assert len(entity_cnt) > 0
+ # common_entity = entity_cnt.most_common(1)[0][0]
+ # context = f'{" | ".join(keywords)}\n\ncontext: user: tell me something about {common_entity}. system: '
+
+ context = f'{" | ".join(keywords)}\n\ncontext: system: '
+ else:
+ triples = [f"[{triple[0]}][{triple[1]}][{triple[2]}]" for triple in triples]
+ context = f'{";".join(triples)}\n\ncontext: system: '
+
+ data.append(json.dumps({'triples': context, 'text': sample['utterance']}, ensure_ascii=False)+'\n')
+
+ file_name = os.path.join(data_dir, f"{data_split}.json")
+ with open(file_name, "w", encoding='utf-8') as f:
+ f.writelines(data)
+ data_by_split[data_split] = data
+ return data_by_split
+
+
+def get_max_len(data_by_split, tokenizer):
+ for data_split in data_by_split.keys():
+ seq_len = {}
+ for line in data_by_split[data_split]:
+ item = json.loads(line.strip())
+ for column, seq in item.items():
+ seq_len.setdefault(column, [])
+ seq_len[column].append(len(tokenizer.tokenize(seq)))
+ print(f"data split: {data_split}")
+ for column, lens in seq_len.items():
+ print(f'\t{column}\tmax_len: {max(lens)}\tmean_len: {round(sum(lens)/len(lens),2)}')
+
+
+if __name__ == '__main__':
+ from argparse import ArgumentParser
+ parser = ArgumentParser(description="create data for seq2seq training")
+ parser.add_argument('--tasks', '-t', metavar='task_name', nargs='*', choices=['nlg', 'dart'], help='names of tasks')
+ parser.add_argument('--datasets', '-d', metavar='dataset_name', nargs='*', help='names of unified datasets')
+ parser.add_argument('--speaker', '-s', type=str, choices=['user', 'system', 'all'], help='speaker(s)')
+ parser.add_argument('--context_window_size', '-c', type=int, default=0, help='how many contextual utterances are considered')
+ parser.add_argument('--len_tokenizer', '-l', type=str, default=None, help='name or path of tokenizer that used to get seq len')
+ parser.add_argument('--ratio', '-r', type=float, default=None, help='how many data is used for training and evaluation')
+ parser.add_argument('--dial_ids_order', '-o', type=int, default=None, help='which data order is used for experiments')
+ parser.add_argument('--key2gen', '-k', action='store_true', default=False, help='generate data for key2gen models')
+ args = parser.parse_args()
+ print(args)
+ if args.len_tokenizer:
+ tokenizer = AutoTokenizer.from_pretrained(args.len_tokenizer)
+ for dataset_name in tqdm(args.datasets, desc='datasets'):
+ dataset = load_dataset(dataset_name, args.dial_ids_order)
+ if args.ratio:
+ dataset['train'] = dataset['train'][:round(len(dataset['train'])*args.ratio)]
+ dataset['validation'] = dataset['validation'][:round(len(dataset['validation'])*args.ratio)]
+ for task_name in tqdm(args.tasks, desc='tasks', leave=False):
+ if args.key2gen:
+ data_dir = os.path.join('data', task_name, "key2gen_"+(dataset_name if not args.ratio else f'{dataset_name}_{args.ratio}_order{args.dial_ids_order}'))
+ else:
+ data_dir = os.path.join('data', task_name, (dataset_name if not args.ratio else f'{dataset_name}_{args.ratio}_order{args.dial_ids_order}'))
+ data_by_split = eval(f"create_{task_name}_data")(dataset, data_dir, args)
+ if args.len_tokenizer:
+ get_max_len(data_by_split, tokenizer)
diff --git a/convlab2/base_models/t5/key2gen/run_dart.sh b/convlab2/base_models/t5/key2gen/run_dart.sh
new file mode 100644
index 0000000000000000000000000000000000000000..6c346408c18c3ead330018cae58effd97fb45265
--- /dev/null
+++ b/convlab2/base_models/t5/key2gen/run_dart.sh
@@ -0,0 +1,85 @@
+n_gpus=1
+task_name="dart"
+dataset_name="dart"
+speaker="system"
+context_window_size=0
+data_dir="data/${task_name}/${dataset_name}"
+output_dir="output/${task_name}/${dataset_name}"
+cache_dir="../cache"
+logging_dir="${output_dir}/runs"
+train_file="${data_dir}/train.json"
+validation_file="${data_dir}/validation.json"
+test_file="${data_dir}/test.json"
+metric_name_or_path="../nlg/nlg_metric.py"
+metric_for_best_model="bleu"
+source_column="triples"
+target_column="text"
+truncation_side="left"
+max_source_length=512
+max_target_length=512
+model_name_or_path="t5-small"
+per_device_train_batch_size=128
+per_device_eval_batch_size=64
+gradient_accumulation_steps=4
+lr=1e-3
+num_train_epochs=10
+
+python create_data_key2gen.py -t ${task_name} -d ${dataset_name} -s ${speaker} -c ${context_window_size}
+
+python ../run_seq2seq.py \
+ --task_name ${task_name} \
+ --train_file ${train_file} \
+ --validation_file ${validation_file} \
+ --source_column ${source_column} \
+ --target_column ${target_column} \
+ --max_source_length ${max_source_length} \
+ --max_target_length ${max_target_length} \
+ --truncation_side ${truncation_side} \
+ --model_name_or_path ${model_name_or_path} \
+ --do_train \
+ --do_eval \
+ --save_strategy epoch \
+ --evaluation_strategy epoch \
+ --save_total_limit 3 \
+ --prediction_loss_only \
+ --cache_dir ${cache_dir} \
+ --output_dir ${output_dir} \
+ --logging_dir ${logging_dir} \
+ --overwrite_output_dir \
+ --preprocessing_num_workers 4 \
+ --per_device_train_batch_size ${per_device_train_batch_size} \
+ --per_device_eval_batch_size ${per_device_eval_batch_size} \
+ --gradient_accumulation_steps ${gradient_accumulation_steps} \
+ --learning_rate ${lr} \
+ --num_train_epochs ${num_train_epochs} \
+ --adafactor \
+ --gradient_checkpointing
+
+python ../run_seq2seq.py \
+ --task_name ${task_name} \
+ --test_file ${test_file} \
+ --source_column ${source_column} \
+ --target_column ${target_column} \
+ --max_source_length ${max_source_length} \
+ --max_target_length ${max_target_length} \
+ --truncation_side ${truncation_side} \
+ --model_name_or_path ${output_dir} \
+ --do_predict \
+ --predict_with_generate \
+ --metric_name_or_path ${metric_name_or_path} \
+ --cache_dir ${cache_dir} \
+ --output_dir ${output_dir} \
+ --logging_dir ${logging_dir} \
+ --overwrite_output_dir \
+ --preprocessing_num_workers 4 \
+ --per_device_train_batch_size ${per_device_train_batch_size} \
+ --per_device_eval_batch_size ${per_device_eval_batch_size} \
+ --gradient_accumulation_steps ${gradient_accumulation_steps} \
+ --learning_rate ${lr} \
+ --num_train_epochs ${num_train_epochs} \
+ --adafactor \
+ --gradient_checkpointing
+
+python ../nlg/merge_predict_res.py -d ${dataset_name} -s ${speaker} -c ${context_window_size} -p ${output_dir}/generated_predictions.json
+
+python ../../../nlg/evaluate_unified_datasets.py -p ${output_dir}/predictions.json --dataset_name ${dataset_name}
diff --git a/convlab2/base_models/t5/key2gen/run_dart_key2gen.sh b/convlab2/base_models/t5/key2gen/run_dart_key2gen.sh
new file mode 100644
index 0000000000000000000000000000000000000000..8f539f469d58ff169ad7a6c71f0a38a894acf566
--- /dev/null
+++ b/convlab2/base_models/t5/key2gen/run_dart_key2gen.sh
@@ -0,0 +1,85 @@
+n_gpus=1
+task_name="dart"
+dataset_name="dart"
+speaker="system"
+context_window_size=0
+data_dir="data/${task_name}/key2gen_${dataset_name}"
+output_dir="output/${task_name}/key2gen_${dataset_name}"
+cache_dir="../cache"
+logging_dir="${output_dir}/runs"
+train_file="${data_dir}/train.json"
+validation_file="${data_dir}/validation.json"
+test_file="${data_dir}/test.json"
+metric_name_or_path="../nlg/nlg_metric.py"
+metric_for_best_model="bleu"
+source_column="triples"
+target_column="text"
+truncation_side="left"
+max_source_length=512
+max_target_length=512
+model_name_or_path="t5-small"
+per_device_train_batch_size=128
+per_device_eval_batch_size=64
+gradient_accumulation_steps=4
+lr=1e-3
+num_train_epochs=10
+
+python create_data_key2gen.py -t ${task_name} -d ${dataset_name} -s ${speaker} -c ${context_window_size} --key2gen
+
+python ../run_seq2seq.py \
+ --task_name ${task_name} \
+ --train_file ${train_file} \
+ --validation_file ${validation_file} \
+ --source_column ${source_column} \
+ --target_column ${target_column} \
+ --max_source_length ${max_source_length} \
+ --max_target_length ${max_target_length} \
+ --truncation_side ${truncation_side} \
+ --model_name_or_path ${model_name_or_path} \
+ --do_train \
+ --do_eval \
+ --save_strategy epoch \
+ --evaluation_strategy epoch \
+ --save_total_limit 3 \
+ --prediction_loss_only \
+ --cache_dir ${cache_dir} \
+ --output_dir ${output_dir} \
+ --logging_dir ${logging_dir} \
+ --overwrite_output_dir \
+ --preprocessing_num_workers 4 \
+ --per_device_train_batch_size ${per_device_train_batch_size} \
+ --per_device_eval_batch_size ${per_device_eval_batch_size} \
+ --gradient_accumulation_steps ${gradient_accumulation_steps} \
+ --learning_rate ${lr} \
+ --num_train_epochs ${num_train_epochs} \
+ --adafactor \
+ --gradient_checkpointing
+
+python ../run_seq2seq.py \
+ --task_name ${task_name} \
+ --test_file ${test_file} \
+ --source_column ${source_column} \
+ --target_column ${target_column} \
+ --max_source_length ${max_source_length} \
+ --max_target_length ${max_target_length} \
+ --truncation_side ${truncation_side} \
+ --model_name_or_path ${output_dir} \
+ --do_predict \
+ --predict_with_generate \
+ --metric_name_or_path ${metric_name_or_path} \
+ --cache_dir ${cache_dir} \
+ --output_dir ${output_dir} \
+ --logging_dir ${logging_dir} \
+ --overwrite_output_dir \
+ --preprocessing_num_workers 4 \
+ --per_device_train_batch_size ${per_device_train_batch_size} \
+ --per_device_eval_batch_size ${per_device_eval_batch_size} \
+ --gradient_accumulation_steps ${gradient_accumulation_steps} \
+ --learning_rate ${lr} \
+ --num_train_epochs ${num_train_epochs} \
+ --adafactor \
+ --gradient_checkpointing
+
+python ../nlg/merge_predict_res.py -d ${dataset_name} -s ${speaker} -c ${context_window_size} -p ${output_dir}/generated_predictions.json
+
+python ../../../nlg/evaluate_unified_datasets.py -p ${output_dir}/predictions.json --dataset_name ${dataset_name}
diff --git a/convlab2/base_models/t5/key2gen/run_nlg.sh b/convlab2/base_models/t5/key2gen/run_nlg.sh
new file mode 100644
index 0000000000000000000000000000000000000000..65140ce334b4b7e7de1201eeef628ad017d03165
--- /dev/null
+++ b/convlab2/base_models/t5/key2gen/run_nlg.sh
@@ -0,0 +1,85 @@
+n_gpus=1
+task_name="nlg"
+dataset_name=$1
+speaker="system"
+context_window_size=$2
+data_dir="data/${task_name}/${dataset_name}/${speaker}/context_${context_window_size}"
+output_dir="output/${task_name}/${dataset_name}/${speaker}/context_${context_window_size}"
+cache_dir="../cache"
+logging_dir="${output_dir}/runs"
+train_file="${data_dir}/train.json"
+validation_file="${data_dir}/validation.json"
+test_file="${data_dir}/test.json"
+metric_name_or_path="../nlg/nlg_metric.py"
+metric_for_best_model="bleu"
+source_column="context+da"
+target_column="response"
+truncation_side="left"
+max_source_length=512
+max_target_length=512
+model_name_or_path="t5-small"
+per_device_train_batch_size=128
+per_device_eval_batch_size=64
+gradient_accumulation_steps=4
+lr=1e-3
+num_train_epochs=10
+
+python create_data_key2gen.py -t ${task_name} -d ${dataset_name} -s ${speaker} -c ${context_window_size}
+
+python ../run_seq2seq.py \
+ --task_name ${task_name} \
+ --train_file ${train_file} \
+ --validation_file ${validation_file} \
+ --source_column ${source_column} \
+ --target_column ${target_column} \
+ --max_source_length ${max_source_length} \
+ --max_target_length ${max_target_length} \
+ --truncation_side ${truncation_side} \
+ --model_name_or_path ${model_name_or_path} \
+ --do_train \
+ --do_eval \
+ --save_strategy epoch \
+ --evaluation_strategy epoch \
+ --save_total_limit 3 \
+ --prediction_loss_only \
+ --cache_dir ${cache_dir} \
+ --output_dir ${output_dir} \
+ --logging_dir ${logging_dir} \
+ --overwrite_output_dir \
+ --preprocessing_num_workers 4 \
+ --per_device_train_batch_size ${per_device_train_batch_size} \
+ --per_device_eval_batch_size ${per_device_eval_batch_size} \
+ --gradient_accumulation_steps ${gradient_accumulation_steps} \
+ --learning_rate ${lr} \
+ --num_train_epochs ${num_train_epochs} \
+ --adafactor \
+ --gradient_checkpointing
+
+python ../run_seq2seq.py \
+ --task_name ${task_name} \
+ --test_file ${test_file} \
+ --source_column ${source_column} \
+ --target_column ${target_column} \
+ --max_source_length ${max_source_length} \
+ --max_target_length ${max_target_length} \
+ --truncation_side ${truncation_side} \
+ --model_name_or_path ${output_dir} \
+ --do_predict \
+ --predict_with_generate \
+ --metric_name_or_path ${metric_name_or_path} \
+ --cache_dir ${cache_dir} \
+ --output_dir ${output_dir} \
+ --logging_dir ${logging_dir} \
+ --overwrite_output_dir \
+ --preprocessing_num_workers 4 \
+ --per_device_train_batch_size ${per_device_train_batch_size} \
+ --per_device_eval_batch_size ${per_device_eval_batch_size} \
+ --gradient_accumulation_steps ${gradient_accumulation_steps} \
+ --learning_rate ${lr} \
+ --num_train_epochs ${num_train_epochs} \
+ --adafactor \
+ --gradient_checkpointing
+
+python ../nlg/merge_predict_res.py -d ${dataset_name} -s ${speaker} -c ${context_window_size} -p ${output_dir}/generated_predictions.json
+
+python ../../../nlg/evaluate_unified_datasets.py -p ${output_dir}/predictions.json --dataset_name ${dataset_name}
diff --git a/convlab2/base_models/t5/key2gen/run_nlg_fewshot.sh b/convlab2/base_models/t5/key2gen/run_nlg_fewshot.sh
new file mode 100644
index 0000000000000000000000000000000000000000..b831beef20ca0d44eb8809c8823d2fdb56b656f7
--- /dev/null
+++ b/convlab2/base_models/t5/key2gen/run_nlg_fewshot.sh
@@ -0,0 +1,88 @@
+n_gpus=1
+task_name="nlg"
+dataset_name=$1
+speaker="system"
+context_window_size=$2
+ratio=$3
+dial_ids_order=$4
+data_dir="data/${task_name}/${dataset_name}_${ratio}_order${dial_ids_order}/${speaker}/context_${context_window_size}"
+output_dir="output/${task_name}/${dataset_name}_${ratio}_order${dial_ids_order}/${speaker}/context_${context_window_size}"
+cache_dir="../cache"
+logging_dir="${output_dir}/runs"
+train_file="${data_dir}/train.json"
+validation_file="${data_dir}/validation.json"
+test_file="${data_dir}/test.json"
+metric_name_or_path="../nlg/nlg_metric.py"
+metric_for_best_model="bleu"
+source_column="context+da"
+target_column="response"
+truncation_side="left"
+max_source_length=512
+max_target_length=512
+model_name_or_path="t5-small"
+per_device_train_batch_size=128
+per_device_eval_batch_size=64
+gradient_accumulation_steps=4
+lr=1e-3
+num_train_epochs=100
+
+python create_data_key2gen.py -t ${task_name} -d ${dataset_name} -s ${speaker} -c ${context_window_size} -r ${ratio} -o ${dial_ids_order}
+
+python ../run_seq2seq.py \
+ --task_name ${task_name} \
+ --train_file ${train_file} \
+ --validation_file ${validation_file} \
+ --source_column ${source_column} \
+ --target_column ${target_column} \
+ --max_source_length ${max_source_length} \
+ --max_target_length ${max_target_length} \
+ --truncation_side ${truncation_side} \
+ --model_name_or_path ${model_name_or_path} \
+ --do_train \
+ --do_eval \
+ --save_strategy epoch \
+ --evaluation_strategy epoch \
+ --save_total_limit 3 \
+ --prediction_loss_only \
+ --load_best_model_at_end \
+ --cache_dir ${cache_dir} \
+ --output_dir ${output_dir} \
+ --logging_dir ${logging_dir} \
+ --overwrite_output_dir \
+ --preprocessing_num_workers 4 \
+ --per_device_train_batch_size ${per_device_train_batch_size} \
+ --per_device_eval_batch_size ${per_device_eval_batch_size} \
+ --gradient_accumulation_steps ${gradient_accumulation_steps} \
+ --learning_rate ${lr} \
+ --num_train_epochs ${num_train_epochs} \
+ --adafactor \
+ --gradient_checkpointing
+
+python ../run_seq2seq.py \
+ --task_name ${task_name} \
+ --test_file ${test_file} \
+ --source_column ${source_column} \
+ --target_column ${target_column} \
+ --max_source_length ${max_source_length} \
+ --max_target_length ${max_target_length} \
+ --truncation_side ${truncation_side} \
+ --model_name_or_path ${output_dir} \
+ --do_predict \
+ --predict_with_generate \
+ --metric_name_or_path ${metric_name_or_path} \
+ --cache_dir ${cache_dir} \
+ --output_dir ${output_dir} \
+ --logging_dir ${logging_dir} \
+ --overwrite_output_dir \
+ --preprocessing_num_workers 4 \
+ --per_device_train_batch_size ${per_device_train_batch_size} \
+ --per_device_eval_batch_size ${per_device_eval_batch_size} \
+ --gradient_accumulation_steps ${gradient_accumulation_steps} \
+ --learning_rate ${lr} \
+ --num_train_epochs ${num_train_epochs} \
+ --adafactor \
+ --gradient_checkpointing
+
+python ../nlg/merge_predict_res.py -d ${dataset_name} -s ${speaker} -c ${context_window_size} -p ${output_dir}/generated_predictions.json -o ${dial_ids_order}
+
+python ../../../nlg/evaluate_unified_datasets.py -p ${output_dir}/predictions.json --dataset_name ${dataset_name}
diff --git a/convlab2/base_models/t5/key2gen/run_nlg_fewshot_key2gen.sh b/convlab2/base_models/t5/key2gen/run_nlg_fewshot_key2gen.sh
new file mode 100644
index 0000000000000000000000000000000000000000..1ffcc869e554e6002cb4408e7eb3607acb9f1119
--- /dev/null
+++ b/convlab2/base_models/t5/key2gen/run_nlg_fewshot_key2gen.sh
@@ -0,0 +1,88 @@
+n_gpus=1
+task_name="nlg"
+dataset_name=$1
+speaker="system"
+context_window_size=$2
+ratio=$3
+dial_ids_order=$4
+data_dir="data/${task_name}/key2gen_${dataset_name}_${ratio}_order${dial_ids_order}/${speaker}/context_${context_window_size}"
+output_dir="output/${task_name}/key2gen_${dataset_name}_${ratio}_order${dial_ids_order}/${speaker}/context_${context_window_size}"
+cache_dir="../cache"
+logging_dir="${output_dir}/runs"
+train_file="${data_dir}/train.json"
+validation_file="${data_dir}/validation.json"
+test_file="${data_dir}/test.json"
+metric_name_or_path="../nlg/nlg_metric.py"
+metric_for_best_model="bleu"
+source_column="context+da"
+target_column="response"
+truncation_side="left"
+max_source_length=512
+max_target_length=512
+model_name_or_path="t5-small"
+per_device_train_batch_size=128
+per_device_eval_batch_size=64
+gradient_accumulation_steps=4
+lr=1e-3
+num_train_epochs=100
+
+python create_data_key2gen.py -t ${task_name} -d ${dataset_name} -s ${speaker} -c ${context_window_size} -r ${ratio} -o ${dial_ids_order} --key2gen
+
+python ../run_seq2seq.py \
+ --task_name ${task_name} \
+ --train_file ${train_file} \
+ --validation_file ${validation_file} \
+ --source_column ${source_column} \
+ --target_column ${target_column} \
+ --max_source_length ${max_source_length} \
+ --max_target_length ${max_target_length} \
+ --truncation_side ${truncation_side} \
+ --model_name_or_path ${model_name_or_path} \
+ --do_train \
+ --do_eval \
+ --save_strategy epoch \
+ --evaluation_strategy epoch \
+ --save_total_limit 3 \
+ --prediction_loss_only \
+ --load_best_model_at_end \
+ --cache_dir ${cache_dir} \
+ --output_dir ${output_dir} \
+ --logging_dir ${logging_dir} \
+ --overwrite_output_dir \
+ --preprocessing_num_workers 4 \
+ --per_device_train_batch_size ${per_device_train_batch_size} \
+ --per_device_eval_batch_size ${per_device_eval_batch_size} \
+ --gradient_accumulation_steps ${gradient_accumulation_steps} \
+ --learning_rate ${lr} \
+ --num_train_epochs ${num_train_epochs} \
+ --adafactor \
+ --gradient_checkpointing
+
+python ../run_seq2seq.py \
+ --task_name ${task_name} \
+ --test_file ${test_file} \
+ --source_column ${source_column} \
+ --target_column ${target_column} \
+ --max_source_length ${max_source_length} \
+ --max_target_length ${max_target_length} \
+ --truncation_side ${truncation_side} \
+ --model_name_or_path ${output_dir} \
+ --do_predict \
+ --predict_with_generate \
+ --metric_name_or_path ${metric_name_or_path} \
+ --cache_dir ${cache_dir} \
+ --output_dir ${output_dir} \
+ --logging_dir ${logging_dir} \
+ --overwrite_output_dir \
+ --preprocessing_num_workers 4 \
+ --per_device_train_batch_size ${per_device_train_batch_size} \
+ --per_device_eval_batch_size ${per_device_eval_batch_size} \
+ --gradient_accumulation_steps ${gradient_accumulation_steps} \
+ --learning_rate ${lr} \
+ --num_train_epochs ${num_train_epochs} \
+ --adafactor \
+ --gradient_checkpointing
+
+python ../nlg/merge_predict_res.py -d ${dataset_name} -s ${speaker} -c ${context_window_size} -p ${output_dir}/generated_predictions.json -o ${dial_ids_order}
+
+python ../../../nlg/evaluate_unified_datasets.py -p ${output_dir}/predictions.json --dataset_name ${dataset_name}
diff --git a/convlab2/base_models/t5/key2gen/run_nlg_key2gen.sh b/convlab2/base_models/t5/key2gen/run_nlg_key2gen.sh
new file mode 100644
index 0000000000000000000000000000000000000000..38807e9a97027a636a7cd54a552f60283fceda0a
--- /dev/null
+++ b/convlab2/base_models/t5/key2gen/run_nlg_key2gen.sh
@@ -0,0 +1,85 @@
+n_gpus=1
+task_name="nlg"
+dataset_name=$1
+speaker="system"
+context_window_size=$2
+data_dir="data/${task_name}/key2gen_${dataset_name}/${speaker}/context_${context_window_size}"
+output_dir="output/${task_name}/key2gen_${dataset_name}/${speaker}/context_${context_window_size}"
+cache_dir="../cache"
+logging_dir="${output_dir}/runs"
+train_file="${data_dir}/train.json"
+validation_file="${data_dir}/validation.json"
+test_file="${data_dir}/test.json"
+metric_name_or_path="../nlg/nlg_metric.py"
+metric_for_best_model="bleu"
+source_column="context+da"
+target_column="response"
+truncation_side="left"
+max_source_length=512
+max_target_length=512
+model_name_or_path="t5-small"
+per_device_train_batch_size=128
+per_device_eval_batch_size=64
+gradient_accumulation_steps=4
+lr=1e-3
+num_train_epochs=10
+
+python create_data_key2gen.py -t ${task_name} -d ${dataset_name} -s ${speaker} -c ${context_window_size} --key2gen
+
+python ../run_seq2seq.py \
+ --task_name ${task_name} \
+ --train_file ${train_file} \
+ --validation_file ${validation_file} \
+ --source_column ${source_column} \
+ --target_column ${target_column} \
+ --max_source_length ${max_source_length} \
+ --max_target_length ${max_target_length} \
+ --truncation_side ${truncation_side} \
+ --model_name_or_path ${model_name_or_path} \
+ --do_train \
+ --do_eval \
+ --save_strategy epoch \
+ --evaluation_strategy epoch \
+ --save_total_limit 3 \
+ --prediction_loss_only \
+ --cache_dir ${cache_dir} \
+ --output_dir ${output_dir} \
+ --logging_dir ${logging_dir} \
+ --overwrite_output_dir \
+ --preprocessing_num_workers 4 \
+ --per_device_train_batch_size ${per_device_train_batch_size} \
+ --per_device_eval_batch_size ${per_device_eval_batch_size} \
+ --gradient_accumulation_steps ${gradient_accumulation_steps} \
+ --learning_rate ${lr} \
+ --num_train_epochs ${num_train_epochs} \
+ --adafactor \
+ --gradient_checkpointing
+
+python ../run_seq2seq.py \
+ --task_name ${task_name} \
+ --test_file ${test_file} \
+ --source_column ${source_column} \
+ --target_column ${target_column} \
+ --max_source_length ${max_source_length} \
+ --max_target_length ${max_target_length} \
+ --truncation_side ${truncation_side} \
+ --model_name_or_path ${output_dir} \
+ --do_predict \
+ --predict_with_generate \
+ --metric_name_or_path ${metric_name_or_path} \
+ --cache_dir ${cache_dir} \
+ --output_dir ${output_dir} \
+ --logging_dir ${logging_dir} \
+ --overwrite_output_dir \
+ --preprocessing_num_workers 4 \
+ --per_device_train_batch_size ${per_device_train_batch_size} \
+ --per_device_eval_batch_size ${per_device_eval_batch_size} \
+ --gradient_accumulation_steps ${gradient_accumulation_steps} \
+ --learning_rate ${lr} \
+ --num_train_epochs ${num_train_epochs} \
+ --adafactor \
+ --gradient_checkpointing
+
+python ../nlg/merge_predict_res.py -d ${dataset_name} -s ${speaker} -c ${context_window_size} -p ${output_dir}/generated_predictions.json
+
+python ../../../nlg/evaluate_unified_datasets.py -p ${output_dir}/predictions.json --dataset_name ${dataset_name}