Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / yeyupiaoling/Whisper-Finetune issues and pull requests

#92 - 请问大佬训练时报错应该怎么解决？

Issue - State: open - Opened by Alone749-i about 1 month ago

#91 - 大佬请问一下数据集中的 duration字段是什么含义？

Issue - State: open - Opened by Alone749-i about 1 month ago - 1 comment

#90 - torch.cuda.OutOfMemoryError: CUDA out of memory

Issue - State: closed - Opened by farmer21cn 2 months ago - 1 comment

#89 - 如果数据集中有的词不在底模的词表中，会有什么问题？

Issue - State: closed - Opened by ILG2021 2 months ago - 1 comment

#88 - 微调后，音频的语言检测结果不准了。

Issue - State: open - Opened by charleybin 3 months ago - 8 comments

#87 - 使用4090 lora微调速度还是很慢，不知道是什么原因

Issue - State: closed - Opened by ILG2021 3 months ago - 2 comments

#85 - 缺少噪音文件夹问题，dataset/noise

Issue - State: closed - Opened by yddnrg 4 months ago - 1 comment

#84 - whisper-medium微调之后，转换模型失败

Issue - State: closed - Opened by chengminglei77 4 months ago - 1 comment

#82 - whisper medium ct2 int8量化之后十多段话集中在一起输出

Issue - State: closed - Opened by joseph16388 4 months ago - 2 comments

#81 - 增加与 infer_gui.py 不同的录音方式。

Pull Request - State: closed - Opened by WThirteen 4 months ago - 1 comment

#80 - 多卡训练卡死

Issue - State: closed - Opened by qaz159qaz159 5 months ago - 1 comment

#79 - Add push to hub support for fine-tuning script

Pull Request - State: closed - Opened by sanchit-gandhi 5 months ago - 1 comment

#78 - Push to Hub Support

Issue - State: closed - Opened by sanchit-gandhi 5 months ago - 2 comments

#76 - whisper全量微调相关问题

Issue - State: closed - Opened by wangyarududu 6 months ago - 1 comment

#75 - tools/create_wenetspeech_data.py报错

Issue - State: closed - Opened by Larry-Ye 6 months ago - 1 comment

#74 - 是否支持mp4 视频

Issue - State: closed - Opened by lizhanyang505 7 months ago - 1 comment

#73 - Belle-whisper-large-v3-zh 没法输出标点符号

Issue - State: closed - Opened by jianchaozhuang 7 months ago - 1 comment

#71 - 建議您整理一下 requirements.txt 及 readme.md

Issue - State: closed - Opened by bardenthenry 7 months ago - 1 comment

#70 - NaN or Inf found in input tensor

Issue - State: closed - Opened by GoldenLinlin 7 months ago - 2 comments

#69 - 如何导出pt格式的模型？

Issue - State: closed - Opened by jinchao123 7 months ago - 1 comment

#68 - 训练出的模型如何导出npz格式

Issue - State: closed - Opened by jackwenshann 8 months ago - 1 comment

#67 - 使用large-v2-finetune，高機率出現重複內容…

Issue - State: closed - Opened by zero-zen 8 months ago - 2 comments

#66 - ubuntu20.04环境问题

Issue - State: closed - Opened by happylsr 8 months ago - 8 comments

#65 - 能提供fine-tune 模型的原始 checkpoint (pt) 吗？

Issue - State: closed - Opened by walletiger 8 months ago - 1 comment

#64 - transcribe时如何产生时间戳？

Issue - State: closed - Opened by Koyonyong 8 months ago - 2 comments

#63 - 微调模型保存时需要连接huggingface

Issue - State: closed - Opened by qiangduscu 8 months ago - 1 comment

#62 - 能結合 pyannote 嗎?

Issue - State: closed - Opened by HKLee2040 8 months ago - 1 comment

#61 - 请问如何导出onnx

Issue - State: closed - Opened by drilistbox 8 months ago - 1 comment

#59 - 模型微调后出现乱码

Issue - State: closed - Opened by hcguoO0 9 months ago - 1 comment

#58 - [CONTRIBUTION] Speech dataset Generator

Issue - State: closed - Opened by davidmartinrius 9 months ago

#57 - 请问加速推理可以离线使用吗

Issue - State: closed - Opened by 20246688 9 months ago - 1 comment

#56 - evaluate运行时出现问题

Issue - State: closed - Opened by yfwang214 9 months ago - 6 comments

#55 - Fix: whitespace error

Pull Request - State: closed - Opened by ZeeshanGeoPk 9 months ago - 4 comments

#53 - 可以使用initial_prompt做微调吗

Issue - State: closed - Opened by ben-8878 10 months ago - 1 comment

#52 - 使用 ct2 轉換後掉辨識率

Issue - State: closed - Opened by ken19980727 10 months ago - 2 comments

#51 - WhisperForConditionalGeneration 與 AutoModelForSpeechSeq2Seq 差異

Issue - State: closed - Opened by ken19980727 10 months ago - 1 comment

#50 - 可以单独导出我们微调模型训练出的数据吗

Issue - State: closed - Opened by jackwenshann 10 months ago - 2 comments

#49 - 123

Issue - State: closed - Opened by 20246688 10 months ago - 2 comments

#48 - 可以微调新的语种吗？

Issue - State: closed - Opened by kli017 10 months ago - 1 comment

#47 - 如何转换V3版本

Issue - State: closed - Opened by xyx361100238 10 months ago - 4 comments

#46 - 資料格式的 language 設定

Issue - State: closed - Opened by ken19980727 11 months ago - 1 comment

#45 - config.json文件与huggingface上的config.json不一样

Issue - State: closed - Opened by hahazei 11 months ago - 1 comment

#44 - 微调时的奇怪问题，训练集变大之后，准确度反而下降了

Issue - State: closed - Opened by ILG2021 11 months ago - 5 comments

#43 - whisper large v3 Fine-Tune 後變得不太能辨識語音

Issue - State: closed - Opened by bardenthenry 11 months ago - 28 comments

#42 - 微调在WhisperProcessor.from_pretrained调用时就报错

Issue - State: closed - Opened by lichq5 11 months ago - 5 comments

#41 - 如何随机化模型参数，从头开始训练

Issue - State: closed - Opened by xyx361100238 11 months ago - 2 comments

#40 - 2卡训练速度比单卡训练快很多

Issue - State: closed - Opened by pcqiao0 12 months ago - 15 comments

#39 - 正常数据和空数据一起训练的格式

Issue - State: closed - Opened by xyx361100238 12 months ago - 9 comments

#38 - 训练稳定后的loss，大家一般都是多少啊

Issue - State: closed - Opened by zouzoutingting about 1 year ago - 1 comment

#37 - 训练数据里，需要提前把标点符号去掉吗

Issue - State: closed - Opened by zouzoutingting about 1 year ago

#36 - 双通道的数据，还是单通道

Issue - State: closed - Opened by zouzoutingting about 1 year ago

#35 - Keyword arguments {'sampling_rate': 16000} not recognized.

Issue - State: closed - Opened by ken19980727 about 1 year ago - 4 comments

#34 - lora微调large V2版本，需要多大的显存，

Issue - State: closed - Opened by zouzoutingting about 1 year ago - 1 comment

#33 - 一般需要多少数据，可以有不错的效果呢

Issue - State: closed - Opened by zouzoutingting about 1 year ago - 1 comment

#32 - 关于DataCollatorSpeechSeq2SeqWithPadding的一处问题

Issue - State: closed - Opened by ILG2021 about 1 year ago - 4 comments

#31 - "无语音数据训练"是什么训练方式啊

Issue - State: closed - Opened by zouzoutingting about 1 year ago - 1 comment

#30 - 训练过程占用显存过高的问题

Issue - State: closed - Opened by xyx361100238 about 1 year ago - 8 comments

#29 - 使用lora微调时遇到的奇怪问题

Issue - State: closed - Opened by ILG2021 about 1 year ago - 2 comments

#29 - 使用lora微调时遇到的奇怪问题

Issue - State: closed - Opened by ILG2021 about 1 year ago - 2 comments

#28 - 训练发生异常

Issue - State: closed - Opened by ILG2021 about 1 year ago - 7 comments

#26 - ada lora 训练问题

Issue - State: closed - Opened by tongjinle123 about 1 year ago - 1 comment

#25 - 微调之后的模型中还有adalora的参数吗

Issue - State: closed - Opened by snsun about 1 year ago - 3 comments

#25 - 微调之后的模型中还有adalora的参数吗

Issue - State: closed - Opened by snsun about 1 year ago - 3 comments

#24 - LoRA参数

Issue - State: closed - Opened by biofoolgreen about 1 year ago - 4 comments

#24 - LoRA参数

Issue - State: closed - Opened by biofoolgreen about 1 year ago - 4 comments

#22 - 低于16khz的音频能做Whisper训练或微调吗？

Issue - State: closed - Opened by zheng1237 about 1 year ago - 1 comment

#22 - 低于16khz的音频能做Whisper训练或微调吗？

Issue - State: closed - Opened by zheng1237 about 1 year ago - 1 comment

#19 - 請問您微調的模型有百度網盤以外的備份嗎?

Issue - State: closed - Opened by aszk1415 over 1 year ago - 5 comments

#19 - 請問您微調的模型有百度網盤以外的備份嗎?

Issue - State: closed - Opened by aszk1415 over 1 year ago - 5 comments

#18 - 多卡训练爆ram

Issue - State: closed - Opened by qnxgy921 over 1 year ago - 2 comments

#18 - 多卡训练爆ram

Issue - State: closed - Opened by qnxgy921 over 1 year ago - 2 comments

#16 - Error: libcudnn_ops_infer.so.8

Issue - State: closed - Opened by xiaobuguilaile over 1 year ago - 3 comments

#16 - Error: libcudnn_ops_infer.so.8

Issue - State: closed - Opened by xiaobuguilaile over 1 year ago - 3 comments

#15 - 关于whisper微调

Issue - State: closed - Opened by ouyang982020 over 1 year ago - 8 comments

#15 - 关于whisper微调

Issue - State: closed - Opened by ouyang982020 over 1 year ago - 8 comments

#14 - accelerate方式训练是否支持deepspeed？

Issue - State: closed - Opened by Johnson-NLP over 1 year ago - 6 comments

#14 - accelerate方式训练是否支持deepspeed？

Issue - State: closed - Opened by Johnson-NLP over 1 year ago - 6 comments

#13 - Android端的低版本能用吗？

Issue - State: closed - Opened by xclidongbo over 1 year ago - 3 comments

#13 - Android端的低版本能用吗？

Issue - State: closed - Opened by xclidongbo over 1 year ago - 3 comments

#11 - 是否可以直接使用CT2版本的模型来进行微调？

Issue - State: closed - Opened by majijia505 over 1 year ago - 1 comment

#11 - 是否可以直接使用CT2版本的模型来进行微调？

Issue - State: closed - Opened by majijia505 over 1 year ago - 1 comment

#10 - 音频文件没有语音内容，却还是是别出了结果

Issue - State: closed - Opened by Thunderltx over 1 year ago - 3 comments