 Andrew Morgan
|
29199c5b2c
Fix UnboundLocalError for tokenizer in from_pretrained (#1238)
|
2 周之前 |
 A
|
1618b7bea2
fix: replace in-place tensor operations for torch.compile compatibility (#1179)
|
1 月之前 |
 Whale and Dolphin
|
b72bcb3163
S2 beta (#1165)
|
1 月之前 |
 Leng Yue
|
daa9b4f31c
S2 beta (#1164)
|
1 月之前 |
 Nazrul Islam
|
80db25e726
fix: clarify model weight loading log message (#1127)
|
5 月之前 |
 PoTaTo
|
781bf1cd7a
Finetune support of OpenAudio-S1 (#1115)
|
5 月之前 |
 Stardust·减
|
83afe683e4
Fix audio quality bug (#1014)
|
10 月之前 |
|
Lengyue
|
89474bbb3b
support s1 model structure
|
10 月之前 |
 heiway
|
d76b9174d1
fix a dtype mismatch when use mps (#790)
|
1 年之前 |
 Whale and Dolphin
|
40665e1a39
[fix]fix problems to let version 1.5 support sft (#774)
|
1 年之前 |
 spicysama
|
b951de3b72
v1.5 (#696)
|
1 年之前 |
|
PoTaTo
|
834b07257c
Agent inference (#650)
|
1 年之前 |
|
spicysama
|
e37a445f51
Fix backend (#627)
|
1 年之前 |
 bfs18
|
dfc95d0f5c
Add preprocessing and validation steps when loading the checkpoint. (#433)
|
1 年之前 |
|
spicysama
|
ea53678446
Quantization Support (#316)
|
1 年之前 |
|
Lengyue
|
c9209e94d9
support finetuning
|
1 年之前 |
|
spicysama
|
584be1d047
Adaptation to version 1.2 (#301)
|
1 年之前 |
|
Leng Yue
|
5e7914472f
This PR brings V1.2 inference into main (#300)
|
1 年之前 |
|
Lengyue
|
26af12b8c6
Make AR and naive decoder configurable
|
2 年之前 |
|
Lengyue
|
3d002fd18b
Optimize amp training
|
2 年之前 |
|
Lengyue
|
87ca8cf62e
Rollback decoding
|
2 年之前 |
|
Lengyue
|
6e67ebda73
Optimize training speed by skip unused decode
|
2 年之前 |
|
Lengyue
|
0a1986bb14
Fix data leaking
|
2 年之前 |
|
Lengyue
|
362a1a7116
Update dual ar generate & config
|
2 年之前 |
|
Lengyue
|
9e1a9debfd
fix bug & implement generate
|
2 年之前 |
|
Lengyue
|
7a31b4043a
Fix target rotate
|
2 年之前 |
|
Lengyue
|
0855b454e4
Add VITS vq
|
2 年之前 |
|
Lengyue
|
cecee062c4
Remove flash attn deps as PyTorch ships in 2.2.0
|
2 年之前 |
|
Lengyue
|
415c78a7d4
Disable neft on eval
|
2 年之前 |
|
Lengyue
|
f5a2df2d23
Add 4 in 8 out support
|
2 年之前 |
