From c2e4e3c2e9be855277d9f4fa9cd0544892ff829a Mon Sep 17 00:00:00 2001 From: 游雁 <zhifu.gzf@alibaba-inc.com> Date: 星期三, 30 八月 2023 09:57:30 +0800 Subject: [PATCH] Merge branch 'main' of github.com:alibaba-damo-academy/FunASR add --- egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/RESULTS.md | 28 +++++++++++++--------------- 1 files changed, 13 insertions(+), 15 deletions(-) diff --git a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/RESULTS.md b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/RESULTS.md index 5f1abb6..2b7269c 100644 --- a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/RESULTS.md +++ b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/RESULTS.md @@ -72,24 +72,24 @@ ## Fine-tuning - Train config: - Training data: aishell-1 - - Training info: lr 0.0002, batch size 2000, 2 gpu, acc_grad 1, 20 epochs + - Training info: lr 0.0002, dataset_type: small, batch bins 2000, 2 gpu, acc_grad 1, 20 epochs - Decoding info: beam_size 1, average_num 10 | model | dev cer(%) | test cer(%) | |:---------:|:-------------:|:-------------:| | Pretrain | 1.75 |1.95 | -| Finetune | 1.62 |1.78 | +| Full-tuning | 1.62 |1.78 | - Train config: - Training data: 16k sichuan dialect - - Training info: lr 0.0002, batch size 2000, 2 gpu, acc_grad 1, 20 epochs + - Training info: lr 0.0002, dataset_type: small, batch bins 2000, 2 gpu, acc_grad 1, 20 epochs - Decoding info: beam_size 1, average_num 10 -| model | Training Data(h) | cn cer(%) | sichuan cer(%) | +| model | Training Data(h) | common cer(%) | sichuan cer(%) | |:--------:|:-------------:|:-------:|:------------:| | Pretrain | | 8.57 | 19.81 | -| Finetune | 50 | 8.8 | 12 | +| Full-tuning | 50 | 8.8 | 12 | | | 100 | 9.24 | 11.63 | | | 200 | 9.82 | 10.47 | | | 300 | 9.95 | 10.44 | @@ -99,16 +99,14 @@ ## Lora Fine-tuning - Train config: - Training data: 16k sichuan dialect - - Training info: lr 0.0002, batch size 2000, 2 gpu, acc_grad 1, 20 epochs + - Training info: lr 0.0002, dataset_type: small, batch bins 2000, 2 gpu, acc_grad 1, 20 epochs - Lora info: lora_bias: "all", lora_list ['q','v'], lora_rank:8, lora_alpha:16, lora_dropout:0.1 - Decoding info: beam_size 1, average_num 10 -| model | Training Data(h) | Trainable Parameters(M) | cn cer(%) | sichuan cer(%) | -|:-------------:|:----------------:|:-----------------------:|:---------:|:--------------:| -| Pretrain | | | 8.57 | 19.81 | -| | | | | | -| Finetune | 50 | 220.9 | 8.8 | 12 | -| Lora Finetune | 50 | 2.29 | 9.13 | 12.13 | -| | | | | | -| Finetune | 200 | 220.9 | 9.82 | 10.47 | -| Lora Finetune | 200 | 2.29 | 9.21 | 11.28 | +| model | Training Data(h) | Trainable Parameters(M) | Memory Usage(G) | common cer(%) | sichuan cer(%) | +|:---------------:|:------------------:|:-------------------------:|:-----------------:|:---------------:|:----------------:| +| Pretrain | | | | 8.57 | 19.81 | +| Full-tuning | 50 | 220.9 | 15 | 8.8 | 12 | +| Lora Finetune | 50 | 2.29 | 7 | 9.13 | 12.13 | +| Full-tuning | 200 | 220.9 | 15 | 9.82 | 10.47 | +| Lora Finetune | 200 | 2.29 | 7 | 9.21 | 11.28 | -- Gitblit v1.9.1