k2-fsa icefall · Discussions · GitHub

8000 k2-fsa icefall · Discussions · GitHub

More Web Proxy on the site http://driver.im/

Slow training of pruned transducer is normal?
Q&A RuABraun

Sort by: Lat 8000 est activity

Discussions

You must be logged in to vote

Is that possible to use Byte BPE model (BBPE) in SSL hubert ?

redsheep913 asked May 28, 2025 in Q&A · Answered

3
You must be logged in to vote

pruned_transducer_stateless7_streaming model finetuning problem

younghounSon asked May 29, 2025 in Q&A · Closed · Unanswered

0
You must be logged in to vote

ASR project whisper_llm_zh

ADMIN2021020130 asked Apr 8, 2025 in Q&A · Unanswered

1
You must be logged in to vote

Is it possible to train with Ground Truth frame alignment for Zipformer?

chiiyeh asked Oct 30 8000 , 2024 in Q&A · Unanswered

0
You must be logged in to vote

Using CR-CTC to train, decoder results are all start and end character.

masterjade7 asked Oct 22, 2024 in Q&A · Answered

15
You must be logged in to vote

Pretrain decoder with large amount of text data.

juliendespres started Sep 10, 2024 in Ideas

3
You must be logged in to vote

Knowledge distillation from a large zipformer2 to a smaller one

joazoa asked Aug 19, 2023 in Q&A · Unanswered

4
You must be logged in to vote

Can transcript_words.txt file contain only unique and sorted lines ?

ChrystianKacki asked May 4, 2024 in Q&A · Closed · Answered

12
You must be logged in to vote

Repository template for model publishing

daniel-dona asked Apr 26, 2024 in Q&A · Answered

1
You must be logged in to vote

Halving BPE vocab size does not reduce mem usage ???

iggygeek asked Apr 17, 2024 in Q&A · Closed · Answered

2
You must be logged in to vote

Data requirements and recommendations for training large-scaled Zipformer2

bharathraj-v asked Apr 3, 2024 in Q&A · Unanswered

0
You must be logged in to vote

Have you tried to initialize the encoder part from a pretrained model ?

iggygeek asked Mar 26, 2024 in Q&A · Answered

5
You must be logged in to vote

Building multimodal systems

AmirHussein96 started Mar 7, 2024 in Ideas · Closed

2
You must be logged in to vote

Is it possible to use the prepare script to download non-SLR12 corpora?

duhumail started Mar 19, 2024 in General

3
You must be logged in to vote

am-scale and lm-scale for "Simple RNNT" loss smoothing

YuriiMytiai asked Feb 7, 2024 in Q&A · Answered

4
You must be logged in to vote

Finetuning facebook/wav2vec2-xls-r-2b

Tirthankar-iiitb started Jan 12, 2024 in General

7
You must be logged in to vote

Training strategy for Zipformer using fp16 ??

ZQuang2202 asked Jan 14, 2024 in Q&A · Unanswered

2
You must be logged in to vote

Is the zipformer2 supported by ncnn ?

joazoa asked Nov 7, 2023 in Q&A · Answered

2
You must be logged in to vote

Training diverges

RuABraun asked Oct 23, 2023 in Q&A · Unanswered

7
You must be logged in to vote

python Segmentation fault

DeityBoom asked Oct 27, 2023 in Q&A · Unanswered

0
You must be logged in to vote

Shallow fusion or rescoring with word-level n-grams for modified_beam_search

mrsrikanth asked Sep 14, 2023 in Q&A · Unanswered

7
You must be logged in to vote

Bypass Temporal Classification: Weakly Supervised Automatic Speech Recognition with Imperfect Transcripts

joazoa asked Sep 12, 2023 in Q&A · Closed · Unanswered

1
You must be logged in to vote

"Open ASR leaderboard"

pehonnet started Sep 8, 2023 in General

1
You must be logged in to vote

Softmax in the self-attention

desh2608 started Aug 30, 2023 in General

4
You must be logged in to vote

Finetuning a model to "forget" all but some sentences

joazoa asked Aug 19, 2023 in Q&A · Unanswered

0

0