OSUM-EChat / conf /data_t2t.yaml
xlgeng's picture
开始部署
841f290
# 文本到文本
#text2text_added_2025_4_4:
# path_huawei: "/mnt/sfs/asr/update_data/text2text_added_by_20250404/shards_list.txt" # 1850
# weight: 1
#text2text_2_added_by_20250409:
# path_huawei: "/mnt/sfs/asr/update_data/text2text_2_added_by_20250409/shards_list.txt" # 2000
# weight: 1
#
#text2text_3_en_added_by_20250411:
# path_huawei: "/mnt/sfs/asr/update_data/text2text_3_en_added_by_20250411/shards_list.txt" # 185
# weight: 1
#
#text2text_4_en_added_by_20250416:
# path_huawei: "/mnt/sfs/asr/update_data/text2text_4_en_added_by_20250416/shards_list.txt"
# weight: 1
#text2text_5_lucy_audioQA_1M_by_cywang_added_by_20250426:
# shard_num: 10000
# path_huawei: "/mnt/sfs/asr/update_data/text2text_5_lucy_audioQA_1M_by_cywang_added_by_20250426/shards_list.txt"
# weight: 0.1
t2t_8772K_by_xlgeng_added_by_20250513:
path: "/home/A03_tmpdata1/text2text_data_xlgeng/t2t_8772K/shards_list.txt"
path_huawei: "/mnt/sfs/asr/update_data/t2t_8772K_by_xlgeng_added_by_20250513/shards_list.txt"
weight: 0.1
#t2t_math_poetry_self_by_xlgeng_added_by_20250513:
# path: "/home/A03_tmpdata1/text2text_data_xlgeng/t2t_math_poetry_self/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/t2t_math_poetry_self_by_xlgeng_added_by_20250513/shards_list.txt" # 75
# weight: 1
Alpaca_CoT_3000W_by_wjt_added_by_20250605:
lab_path_huawei: ""
shard_num: 30000
path_huawei: "/mnt/sfs/asr/update_data/Alpaca_CoT_3000W_by_wjt_added_by_20250605/shards_list.txt"
path: "/home/A03_tmpdata1/text2text_data_xlgeng/Alpaca-CoT_3000W/shards_list.txt"
weight: 0.15
qwenomni_bench_data:
path: "/home/A02_tmpdata3/osum_t2t/qwenomni_bench_data/shards_list.txt"
weight: 3
three_kingdoms:
path: "/home/A02_tmpdata3/osum_t2t/three_kingdoms/shards_list.txt"
weight: 3
voicebench_data:
path: "/home/A02_tmpdata3/osum_t2t/voicebench_data/shards_list.txt"
weight: 3
t2t_osum_self_instruction_8K:
path: "/home/A02_tmpdata3/osum_t2t/t2t_osum_self_instruction_8K/shards_list.txt"
weight: 3
#t2t_osum_self_instruction_8K_by_xlgeng_added_by_20250529:
# path: "/home/A02_tmpdata3/t2t_osum_self_instruction_8K_by_xlgeng_added_by_20250529/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/t2t_osum_self_instruction_8K_by_xlgeng_added_by_20250529/shards_list.txt"
# weight: 5
#kouyu_t2t_data_by_xlgeng_added_by_20250622:
# path: ""
# shard_num: 1758
# path_huawei: "/mnt/sfs/asr/update_data/kouyu_t2t_data_by_xlgeng_added_by_20250622s/shards_list.txt"
# weight: 1
# 4653
#text2text_data_xlgeng_three_kingdoms_by_xlgeng_added_by_20250701:
# path: "/home/A02_tmpdata3/text2text_data_xlgeng_three_kingdoms_by_xlgeng_added_by_20250701/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/text2text_data_xlgeng_three_kingdoms_by_xlgeng_added_by_20250701/shards_list.txt"
# weight: 1
# lab_path_huawei: "/home/work_nfs11/asr_data/data/text2text_data_xlgeng/shard/benchdata/three_kingdoms/shard/shards_list.txt"
# shard_num: 24
#
#text2text_data_xlgeng_qwenomni_bench_data_by_xlgeng_added_by_20250701:
# path: "/home/A02_tmpdata3/text2text_data_xlgeng_qwenomni_bench_data_by_xlgeng_added_by_20250701/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/text2text_data_xlgeng_qwenomni_bench_data_by_xlgeng_added_by_20250701/shards_list.txt"
# weight: 1
# lab_path_huawei: "/home/work_nfs11/asr_data/data/text2text_data_xlgeng/shard/benchdata/qwenomni_bench_data/shard/shards_list.txt"
# shard_num: 113
#text2text_data_xlgeng_voicebench_data_by_xlgeng_added_by_20250701:
# path: "/home/A02_tmpdata3/text2text_data_xlgeng_voicebench_data_by_xlgeng_added_by_20250701/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/text2text_data_xlgeng_voicebench_data_by_xlgeng_added_by_20250701/shards_list.txt"
# weight: 1
# lab_path_huawei: "/home/work_nfs11/asr_data/data/text2text_data_xlgeng/shard/benchdata/voicebench_data/shard/shards_list.txt"
# shard_num: 65
#
#t2t_age_chat_by_cywang_added_by_20250708: # have
# path: "/home/A02_tmpdata3/osum_s2s/t2t_age_chat_by_cywang_added_by_20250708/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/t2t_age_chat_by_cywang_added_by_20250708/shards_list.txt"
# lab_path_huawei: "/home/work_nfs11/asr_data/data/osum_data/t2t_paralanguage_chat/age_chat/shard_dir/shards_list.txt"
# shard_num: 50
# weight: 1
#
#t2t_caption_chat_by_cywang_added_by_20250708: # have
# path: "/home/A02_tmpdata3/osum_s2s/t2t_caption_chat_by_cywang_added_by_20250708/shards_list.txt"
# lab_path_huawei: "/home/work_nfs11/asr_data/data/osum_data/t2t_paralanguage_chat/caption_chat/shard_dir/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/t2t_caption_chat_by_cywang_added_by_20250708/shards_list.txt"
# shard_num: 100
# weight: 1
#
#t2t_emotion_chat_by_cywang_added_by_20250708: # have
# path: "/home/A02_tmpdata3/osum_s2s/t2t_emotion_chat_by_cywang_added_by_20250708/shards_list.txt"
# lab_path_huawei: "/home/work_nfs11/asr_data/data/osum_data/t2t_paralanguage_chat/emotion_chat/shard_dir/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/t2t_emotion_chat_by_cywang_added_by_20250708/shards_list.txt"
# shard_num: 50
# weight: 1
#
#t2t_sex_chat_by_cywang_added_by_20250708: # have
# path: "/home/A02_tmpdata3/osum_s2s/t2t_sex_chat_by_cywang_added_by_20250708/shards_list.txt"
# lab_path_huawei: "/home/work_nfs11/asr_data/data/osum_data/t2t_paralanguage_chat/sex_chat/shard_dir/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/t2t_sex_chat_by_cywang_added_by_20250708/shards_list.txt"
# shard_num: 50
# weight: 1
#
#t2t_xianshi_emotion_chat_by_cywang_added_by_20250711: # no
# path: "/home/A02_tmpdata3/osum_t2t/t2t_xianshi_emotion_chat/shards_list.txt"
# lab_path_huawei: "/home/work_nfs11/asr_data/data/osum_data/t2t_paralanguage_chat/xianshi_emotion_chat/shard_dir/shards_list.txt|/home/work_nfs23/asr_data/data/osum_chat/t2t_data/t2t_paralanguage_chat/xianshi_emotion_chat/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/t2t_xianshi_emotion_chat_by_cywang_added_by_20250711/shards_list.txt"
# shard_num: 50
# weight: 1
#
#t2t_sex_chat_2_by_cywang_added_by_20250711: # no
# path: "/home/A02_tmpdata3/osum_t2t/t2t_sex_chat_2_by_cywang_added_by_20250711/shards_list.txt"
# lab_path_huawei: "/home/work_nfs11/asr_data/data/osum_data/t2t_paralanguage_chat/sex_chat_2/shard_dir/shards_list.txt|/home/work_nfs23/asr_data/data/osum_chat/t2t_data/t2t_paralanguage_chat/t2t_sex_chat_2_by_cywang_added_by_20250711/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/t2t_sex_chat_2_by_cywang_added_by_20250711/shards_list.txt"
# shard_num: 27
# weight: 1
#
#t2t_age_chat_2_by_cywang_added_by_20250711: # no
# path: "/home/A02_tmpdata3/osum_t2t/t2t_age_chat_2_by_cywang_added_by_20250711/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/t2t_age_chat_2_by_cywang_added_by_20250711/shards_list.txt"
# lab_path_huawei: "/home/work_nfs11/asr_data/data/osum_data/t2t_paralanguage_chat/age_chat_2/shard_dir/shards_list.txt|/home/work_nfs23/asr_data/data/osum_chat/t2t_data/t2t_paralanguage_chat/t2t_age_chat_2_by_cywang_added_by_20250711/shards_list.txt"
# shard_num: 27
# weight: 1
#
#t2t_sex_chat_2_by_cywang_added_by_20250715: # no
# path: "/home/A02_tmpdata3/osum_t2t/t2t_sex_chat_2_by_cywang_added_by_20250715/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/t2t_sex_chat_2_by_cywang_added_by_20250715/shards_list.txt"
# lab_path_huawei: "/home/work_nfs14/asr_data/data/osum_data/t2t_paralanguage_chat/sex_chat_2/shard_dir/shards_list.txt"
# shard_num: 10
# weight: 1
#
#t2t_age_chat_3_by_cywang_added_by_20250716: # no
# path: "/home/A02_tmpdata3/osum_t2t/t2t_age_chat_3_by_cywang_added_by_20250716/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/t2t_age_chat_3_by_cywang_added_by_20250716/shards_list.txt"
# lab_path_huawei: "/home/work_nfs14/asr_data/data/osum_data/t2t_paralanguage_chat/age_chat_3/shard_dir/shards_list.txt"
# shard_num: 10
# weight: 1
#
#t2t_caption_chat_3_by_cywang_added_by_20250716: # have
# path: "/home/A02_tmpdata3/osum_t2t/t2t_caption_chat_3_by_cywang_added_by_20250716/shards_list.txt"
# path_huawei: "/mnt/sfs/asr/update_data/t2t_caption_chat_3_by_cywang_added_by_20250716/shards_list.txt"
# lab_path_huawei: "/home/work_nfs14/asr_data/data/osum_data/t2t_paralanguage_chat/caption_chat_3/shard_dir/shards_list.txt"
# shard_num: 10
# weight: 1