May d410a5652d fix(train): align DDP no_sync with gradient accumulation (#3472) ...

Synchronize DDP/FSDP gradients on the final microbatch of each accumulation window.

Verified with a two-rank CPU/Gloo reproduction, py_compile, diff checks, and a clean synthetic merge against current main.

2026-08-05 11:32:24 +08:00

..

auto

fix(auto-model): align split punctuation tokens (#3469)

2026-08-04 14:27:55 +08:00

bin

fix(server): return speaker labels for spk requests (#3433)

2026-07-27 20:01:21 +08:00

datasets

perf: optimize training pipeline for ~3% speedup

2026-05-23 20:56:53 +08:00

download

fix: route paraformer-en to the English HF model (#3466)

2026-08-04 10:06:33 +08:00

frontends

fix: bound streaming VAD frame buffers (#3236)

2026-07-17 17:24:23 +08:00

losses

feat: implement CR-CTC (Consistency-Regularized CTC) loss

2026-05-25 22:45:14 +08:00

metrics

Migrate editdistance metrics to rapidfuzz with empty-reference guard

2026-07-07 15:00:51 +08:00

models

feat(nano): support LoRA finetuning of the Qwen3 LLM (#3456)

2026-08-03 14:28:02 +08:00

optimizers

docs: add docstrings to 1950 functions across 276 files for API documentation

2026-05-21 07:52:52 +00:00

schedulers

docs: add docstrings to 1950 functions across 276 files for API documentation

2026-05-21 07:52:52 +00:00

tokenizer

docs: add docstrings to 1950 functions across 276 files for API documentation

2026-05-21 07:52:52 +00:00

train_utils

fix(train): align DDP no_sync with gradient accumulation (#3472)

2026-08-05 11:32:24 +08:00

utils

Decode container-formatted audio bytes before inference (#3435)

2026-07-27 14:55:49 +00:00

__init__.py

fix: lazy load top-level AutoModel exports

2026-07-19 18:27:49 +08:00

cli.py

Fix CLI SRT sentence segmentation (#3254)

2026-07-19 08:54:44 +08:00

register.py

docs: add docstrings to 1950 functions across 276 files for API documentation

2026-05-21 07:52:52 +00:00

version.txt

chore: prepare FunASR 1.4.1 release

2026-08-04 02:23:14 +00:00