1463 Commits

Author	SHA1	Message	Date
hoshi-hiyouga	fe7057a8a3	Update attention.py	2024-09-29 10:47:41 +08:00
Amirreza A	94ee105526	made a small change to a warning about fa2 for gemma2 models.	2024-09-28 19:03:36 +03:30
hoshi-hiyouga	8e5d12c2c4	add modelscope models	2024-09-26 11:22:48 +08:00
marko1616	885a0b77ab	Chore: Support llama3.2.	2024-09-25 16:08:44 -04:00
hiyouga	ba52103ba7	optionally replace jinja template	2024-09-25 23:02:02 +08:00
hoshi-hiyouga	f230130563	flat string	2024-09-19 16:43:42 +08:00
hoshi-hiyouga	af7f547ecb	lint	2024-09-19 16:21:43 +08:00
hoshi-hiyouga	98b464d2dd	fix bug	2024-09-19 16:21:21 +08:00
hoshi-hiyouga	36611d5c09	improve error message	2024-09-19 16:06:00 +08:00
ybyang	953e1a0fb2	fix: 修复function call数据集如果 function_call 值的为不合法json，异常提示且中断训练。	2024-09-19 15:00:10 +08:00
hoshi-hiyouga	f0b930d94a	fix webui	2024-09-19 02:13:39 +08:00
hoshi-hiyouga	92ef62f502	add qwen2.5 models	2024-09-19 02:07:54 +08:00
Billy Cao	7a2958a44f	Add qwen_vl to liger kernel supported list	2024-09-14 19:28:20 +08:00
hiyouga	0ded765784	set dev version	2024-09-11 18:56:37 +08:00
hiyouga	c7e51ff187	fix #5411	2024-09-11 17:36:42 +08:00
hiyouga	bdde35fd2e	update accelerate ver for schedule_free optimizers	2024-09-09 22:51:08 +08:00
hiyouga	6dc2b00fa2	fix mm plugin	2024-09-09 22:41:28 +08:00
hiyouga	c93795ae14	fix qwen2vl preprocess	2024-09-09 22:33:33 +08:00
hiyouga	90d6df6222	release v0.9.0 (real)	2024-09-09 01:00:25 +08:00
hiyouga	653fe70acb	fix constants	2024-09-08 23:52:30 +08:00
hiyouga	54b5c4b819	release v0.9.0	2024-09-08 23:43:35 +08:00
hiyouga	c9b3870adb	tiny fix	2024-09-08 23:18:08 +08:00
hiyouga	f2aa02c070	update scripts	2024-09-08 14:17:41 +08:00
hiyouga	b6681d7198	support vllm 0.6.0	2024-09-08 02:26:20 +08:00
hiyouga	b332908ab4	fix test case	2024-09-08 01:50:51 +08:00
hiyouga	52a06efaf8	add test case	2024-09-08 01:40:49 +08:00
hiyouga	fb72a3adb0	support activation offloading via unsloth gc	2024-09-08 01:22:19 +08:00
hiyouga	54c6905937	add docstrings, refactor logger	2024-09-08 00:56:56 +08:00
hoshi-hiyouga	36665f3001	fix #5384	2024-09-07 01:21:14 +08:00
hiyouga	76f2e59504	tiny fix	2024-09-05 23:41:16 +08:00
hiyouga	94d5b1bd8f	add e2e tests	2024-09-05 21:52:28 +08:00
hoshi-hiyouga	1274356263	Merge pull request #5372 from LDLINGLINGLING/main ... 增加了对minicpm3.0的适配'	2024-09-05 21:35:42 +08:00
liudan	3d3fbaaff9	根据代码规范修改了代码	2024-09-05 20:17:55 +08:00
hoshi-hiyouga	e9bda48c6d	fix #5366	2024-09-05 18:08:09 +08:00
hiyouga	359ef8bb0e	support Yi-Coder models	2024-09-05 03:12:24 +08:00
hiyouga	1173f7fc1d	fix ci	2024-09-05 03:02:59 +08:00
hiyouga	8cafc7b055	video datasets	2024-09-05 02:04:17 +08:00
liudan	d7ba97be48	增加了对minicpm3.0的适配'	2024-09-04 23:10:05 +08:00
hiyouga	dabad5570b	update get template	2024-09-04 22:36:20 +08:00
hoshi-hiyouga	8f441c2b3a	Merge pull request #5323 from naem1023/feat/add-dataset-map-batch-size-argument ... Add batch size of map function in the preprocessed dataset	2024-09-04 22:09:36 +08:00
hoshi-hiyouga	44d6947e55	fix #5228	2024-09-04 19:10:30 +08:00
hiyouga	d41d43a7c3	fix #5344	2024-09-04 03:06:06 +08:00
hiyouga	47ea97fb1b	lazy image load	2024-09-04 02:27:08 +08:00
hiyouga	59d2b31e96	fix #5334	2024-09-03 19:09:42 +08:00
hiyouga	69d0acacc3	fix #5338	2024-09-03 17:45:17 +08:00
hiyouga	22959bcdd3	lint	2024-09-03 00:46:25 +08:00
hiyouga	a61c8c4890	fix #5324	2024-09-02 23:56:21 +08:00
naem1023	209313eeea	feat: add batch size of map function in the preprocessed dataset	2024-09-02 13:52:47 +09:00
hoshi-hiyouga	99fd9637bd	fix trainer predict	2024-09-02 10:15:29 +08:00
hoshi-hiyouga	a6c6750e8a	remove .cpu()	2024-09-02 10:10:53 +08:00