mirror of https://github.com/k2-fsa/icefall.git synced 2025-08-09 10:02:22 +00:00

History

Fix CI tests. (#1974 )

- Introduce unified AMP helpers (create_grad_scaler, torch_autocast) to handle 
  deprecations in PyTorch ≥2.3.0

- Replace direct uses of torch.cuda.amp.GradScaler and torch.cuda.amp.autocast 
  with the new utilities across all training and inference scripts

- Update all torch.load calls to include weights_only=False for compatibility with 
  newer PyTorch versions

2025-07-01 13:47:55 +08:00

local

Zipformer recipe for CommonVoice (#1546 )

2024-04-09 11:37:08 +08:00

pruned_transducer_stateless7

Fix CI tests. (#1974 )

2025-07-01 13:47:55 +08:00

pruned_transducer_stateless7_streaming

Fix CI tests. (#1974 )

2025-07-01 13:47:55 +08:00

zipformer

Fix CI tests. (#1974 )

2025-07-01 13:47:55 +08:00

prepare.sh

Update prepare.sh (#1768 )

2024-10-09 15:50:12 +08:00

README.md

Zipformer for Common Voice (#997 )

2023-04-17 17:47:25 +08:00

RESULTS.md

Zipformer recipe for CommonVoice (#1546 )

2024-04-09 11:37:08 +08:00

shared

Add Common Voice (#994 )

2023-04-11 20:56:40 +08:00

README.md

Introduction

This recipe includes some different ASR models trained with Common Voice

./RESULTS.md contains the latest results.

Transducers

There are various folders containing the name transducer in this folder. The following table lists the differences among them.

	Encoder	Decoder	Comment
`pruned_transducer_stateless7`	Zipformer	Embedding + Conv1d	First experiment with Zipformer from Dan

The decoder in transducer_stateless is modified from the paper RNN-Transducer with Stateless Prediction Network. We place an additional Conv1d layer right after the input embedding layer.