164 Commits

Author	SHA1	Message	Date
hiyouga	b27269bd2b	add test cases	2024-06-15 04:05:54 +08:00
hiyouga	c94e6c9411	add quant check in webui export tab	2024-06-13 03:19:18 +08:00
hiyouga	6baafd4eb3	fix #4221	2024-06-13 02:48:21 +08:00
hiyouga	cf9f2d6c42	fix #4209 ... DeepSpeed ZeRO3 has inflight param error when calling model.eval()	2024-06-13 02:25:50 +08:00
hiyouga	2ed8270112	clean code	2024-06-13 01:58:16 +08:00
hoshi-hiyouga	1f23f25226	Merge pull request #4246 from hzhaoy/adapt-vllm-v0.5.0 ... adapt vllm==0.5.0	2024-06-13 01:54:02 +08:00
hiyouga	713fde4259	fix lint	2024-06-13 00:48:44 +08:00
hzhaoy	8fb6366ebe	adapt vllm==0.5.0	2024-06-12 18:29:03 +08:00
hiyouga	577de2fa07	fix #4242	2024-06-12 16:50:11 +08:00
Arthur Kim	d65a3f7cb6	Support vllm==0.5.0	2024-06-12 16:49:12 +09:00
hoshi-hiyouga	9049aab911	Merge pull request #4204 from dignfei/main ... fixbug：llama3在增量预训练时应该使用<|end_of_text|>标识文本的结束	2024-06-11 17:06:10 +08:00
hoshi-hiyouga	0c29233237	Update pretrain.py	2024-06-11 17:02:14 +08:00
hiyouga	cca6f35108	fix deepspeed version	2024-06-11 16:52:36 +08:00
d	6979f3f848	经过大量的增量预训练，进行对比试验，发现这个bug：llama3在预训练时使用的tokenizer.eos_toke是'<|end_of_text|>' ，这里在每条数据后面也得用这个，而不是'<|eot_id|>'，否则很容易导致严重的性能下降	2024-06-11 16:23:40 +08:00
hiyouga	89f2bd8c8c	fix #4198	2024-06-11 15:38:38 +08:00
hiyouga	90e14a960d	tiny fix	2024-06-11 12:48:53 +08:00
hiyouga	3f24337a8a	tiny fix	2024-06-11 01:04:16 +08:00
hiyouga	91e62a098f	set dev version	2024-06-11 00:50:53 +08:00
hiyouga	2b6ebd6b51	release v0.8.1	2024-06-11 00:44:26 +08:00
hiyouga	a793e8456b	fix #4160 ... The split heads should be concatenated in dim=2	2024-06-11 00:37:17 +08:00
hiyouga	0012762b04	update evaluator	2024-06-10 23:56:00 +08:00
hiyouga	c907d81667	fix #2666	2024-06-10 21:24:15 +08:00
hiyouga	972ec9c668	fix llamafactory-cli env	2024-06-08 07:15:45 +08:00
hiyouga	3ac11e77cc	set dev version	2024-06-08 06:46:09 +08:00
hiyouga	5aa4ce4756	release v0.8.0	2024-06-08 05:20:54 +08:00
hiyouga	54cd743ebf	reorganize adapter code	2024-06-08 00:47:23 +08:00
hoshi-hiyouga	cfd62283a9	fix #4139	2024-06-08 00:45:02 +08:00
hiyouga	06e5d136a4	add resume args in webui	2024-06-08 00:22:16 +08:00
hiyouga	8bf9da659c	fix #4137	2024-06-07 19:16:06 +08:00
hiyouga	f8d8690bf4	tiny fix	2024-06-07 05:19:21 +08:00
hiyouga	4489d73ac7	fix ppo trainer save zero3 model ... accelerator.get_state_dict(ds_model) should be called at all ranks	2024-06-07 05:14:19 +08:00
hiyouga	2702d7e952	fix ppo in trl 0.8.6	2024-06-07 04:48:29 +08:00
hiyouga	f9e818d79c	fix #4120	2024-06-07 04:18:05 +08:00
hiyouga	ccc8b64cc2	update data processors	2024-06-07 04:15:40 +08:00
hoshi-hiyouga	181dbb0d05	Merge pull request #4009 from AlongWY/main ... supervised packing with greedy knapsack algorithm	2024-06-07 03:48:46 +08:00
hoshi-hiyouga	c09ad8bab3	Update supervised.py	2024-06-07 03:42:08 +08:00
hoshi-hiyouga	788e8232fc	Update supervised.py	2024-06-07 03:38:23 +08:00
hoshi-hiyouga	8cecade708	Update supervised.py	2024-06-07 03:38:04 +08:00
hiyouga	8e95648850	add qwen2 models	2024-06-07 00:22:57 +08:00
hiyouga	74f96efef9	rename files	2024-06-07 00:09:06 +08:00
hiyouga	45d8be8f93	add DISABLE_TORCHRUN option	2024-06-06 23:44:58 +08:00
hoshi-hiyouga	55c18c49b0	Merge pull request #4082 from MengqingCao/bugfix ... Fix #4077	2024-06-06 23:38:40 +08:00
hoshi-hiyouga	751dd77bc0	Update cli.py	2024-06-06 23:38:09 +08:00
hiyouga	76c61905b2	fix ppo+zero3 #3108	2024-06-06 23:30:07 +08:00
hiyouga	451b6693c0	fix torch gc	2024-06-06 20:30:25 +08:00
hiyouga	149610c636	fix ppo dataset bug #4012	2024-06-06 19:03:20 +08:00
hiyouga	fad2591e31	update trainers	2024-06-06 18:45:49 +08:00
hiyouga	67aa78cde0	fix base64 image read #4061	2024-06-06 17:29:19 +08:00
hiyouga	cae4737907	lora modules: all by default	2024-06-06 03:53:28 +08:00
hiyouga	c23cc63d3d	add codestral 22B	2024-06-06 03:42:50 +08:00