Amy

How to intervene the translation result according to the dictionary

1

How to intervene the translation result according to the dictionary？ I have a dictionary, and I want to translate some words into the corresponding professional terms of the dictionary. Does...

[Question]: mbart多对多模型下载后有5G，我在A40 显存44G上面跑不了吗？运行报内存溢出

3

### 请提出你的问题 place = "gpu" paddle.set_device(place) model_name = "mbart-large-50-many-to-many-mmt" tokenizer = MBartTokenizer.from_pretrained(model_name) model = MBartForConditionalGeneration.from_pretrained(model_name, src_lang="en_XX") model.eval() def postprocess_response(seq, bos_idx, eos_idx): """Post-process the decoded sequence.""" eos_pos = len(seq) - 1...

To fine-tune how much gpu is required for the BELLE-7B-2M model, I am now fine-tuning the error memory overflow reported on the a100

6

To fine-tune how much gpu is required for the BELLE-7B-2M model, I am now fine-tuning the error memory overflow reported on the a100

训练mbart模型报错 ，Logits和Label维度不匹配

2

paddlepaddle-gpu 2.5.2.post112 paddlenlp 2.7.1 报错信息： ValueError: (InvalidArgument) Input(Logits) and Input(Label) should in same shape in dimensions except axis. [Hint: Expected logits_dims[i] == labels_dims[i], but received logits_dims[i]:288 != labels_dims[i]:256.] (at ../paddle/phi/infermeta/binary.cc:909)...

部署paddlenlp的mbart模型报错

2

python -m paddle_serving_server.serve --model SERVING_SERVER_DIR --thread 10 --port 9393 --gpu_id 0 执行上面命令报以下错误 Error Message Summary: ---------------------- NotFoundError: Operator (fusion_mbart_decoding) is not registered. [Hint: op_info_ptr should not be null.] (at /paddle/paddle/fluid/framework/op_info.h:151)

对m2m100 和mt5-base模型提供加速，fast transformer支持，动转静支持

1

### Feature request 对m2m100 和mt5-base模型提供加速，fast transformer支持，动转静支持 ### Motivation 对m2m100 和mt5-base模型提供加速，fast transformer支持，动转静支持 ### Your contribution 对m2m100 和mt5-base模型提供加速，fast transformer支持，动转静支持