AmazingJ
AmazingJ
请问我怎么使用自己的数据 continue training? 不是在下游任务,是 pre-training
numpy.core._exceptions._ArrayMemoryError: Unable to allocate 550. GiB for an array with shape (28235788,) and data type |S20921 How did you solve this mistake? The pre training corpus is read incorrectly. My...
希望可以尽快开源 DeBerta continue training 的相关代码。感谢
PAI/nlp_gpt3_text-generation_1.3B_MoE-32 无法使用
RT 例如这样: asr_pipeline = pipeline( task=Tasks.auto_speech_recognition, model="damo/speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404", vad_model='damo/speech_fsmn_vad_zh-cn-16k-common-pytorch', punc_model='damo/punc_ct-transformer_cn-en-common-vocab471067-large', ngpu=1, batch_size=5000) param_dict = dict() param_dict['hotword'] = "短视频" param_dict['clas_scale'] = 1.50 asr_pipeline(_samples, param_dict) 
我使用了 Adv 标准的 benchmark 数据集,为什么跑不到您汇报的 0.08BLEU 呢? export CUDA_VISIBLE_DEVICES=1,2,3,4,5,6,7 python train_t5_pointer_network.py \ --train_file /data/generation/data/AdvertiseGen/t5_train.json \ --dev_file /data/generation/data/AdvertiseGen/t5_dev.json \ --batch_size 64 \ --max_epochs 20 \ --max_source_length 64 \ --max_target_length 256 \...
RT 我使用 FinGPTV3 在中文上进行使用全是乱码,可以提供下代码吗? 感谢