Commit Graph

144 Commits

Author SHA1 Message Date
hiyouga
34f16cc635 follow #4878 fix #4684
Former-commit-id: 779aae83d2
2024-07-18 22:06:12 +08:00
Shiyu Zhang
249adacc4d 仅仅训练最后一轮对话
Former-commit-id: 1e7b396ff2
2024-07-18 15:30:25 +08:00
hoshi-hiyouga
2b22a7da48 Merge pull request #4691 from codemayq/feature-suppot-eval-dataset
add eval dataset support

Former-commit-id: 15b399a82f
2024-07-15 01:00:34 +08:00
hoshi-hiyouga
30a3c6e886 Update preprocess.py
Former-commit-id: df52fb05b1
2024-07-15 00:55:36 +08:00
hoshi-hiyouga
eed7cbb453 Update parser.py
Former-commit-id: 84e4047f8a
2024-07-15 00:55:21 +08:00
hoshi-hiyouga
5633c0ab1e Update data_utils.py
Former-commit-id: 97a0e291c7
2024-07-15 00:54:34 +08:00
hoshi-hiyouga
2e9c9471da Update loader.py
Former-commit-id: a5b809516e
2024-07-15 00:50:06 +08:00
hoshi-hiyouga
140b512426 Update parser.py
Former-commit-id: 3d39d74003
2024-07-14 23:04:34 +08:00
hiyouga
0b26011181 fix gemma2 attention
Former-commit-id: 2f6af73da2
2024-07-13 23:33:45 +08:00
hiyouga
3baf1d1f7e fix #4792
Former-commit-id: 6b48308ef9
2024-07-13 22:07:58 +08:00
hiyouga
7fcffb860d add codegeex4, internlm2.5
Former-commit-id: 53b1002fb7
2024-07-06 16:16:47 +08:00
codingma
74f0d02eb8 1. add custom eval dataset support
2. merge load dataset and split dataset function


Former-commit-id: 76f3bbcfc0
2024-07-05 15:52:10 +08:00
hiyouga
8379a39776 fix processors
Former-commit-id: 9f33f1edf5
2024-07-05 08:33:22 +08:00
hiyouga
9aa3403687 fix #4683
Former-commit-id: e43809bced
2024-07-05 00:58:05 +08:00
hzhaoy
6d892dbc23 tiny fix
Former-commit-id: 738df47748
2024-07-04 10:20:28 +08:00
hiyouga
cbb93a2b47 tiny fix
Former-commit-id: 44747cebd2
2024-07-04 03:02:23 +08:00
hiyouga
4987aa32ba fix data map for packing
Former-commit-id: b5d101e1bf
2024-07-04 03:01:31 +08:00
hiyouga
7b3c1f29ff fix packing for eager/sdpa attn
Former-commit-id: 6fd6aa4530
2024-07-04 01:52:43 +08:00
hoshi-hiyouga
a38ff842d0 Merge pull request #4224 from chuan298/main
Implement efficient packing without cross-contamination attention

Former-commit-id: 87d9b2d005
2024-07-04 01:18:54 +08:00
hiyouga
bfdaadcc40 update packing
Former-commit-id: cce7083024
2024-07-04 01:10:55 +08:00
hiyouga
ff6fc666c1 update hparams
Former-commit-id: 575a02a23d
2024-07-03 23:18:58 +08:00
hiyouga
cc31014002 improve rlhf
Former-commit-id: c47ab6c072
2024-07-02 22:23:08 +08:00
ancv
7f42932957 move efficient_packing from data_args to model_args
Former-commit-id: e8e13b0942
2024-07-02 18:37:55 +07:00
hoshi-hiyouga
2452f57cd7 Merge branch 'main' into main
Former-commit-id: e8e6af2651
2024-07-01 21:01:09 +08:00
hiyouga
ca7b65439d fix #4402 #4617
Deprecate reserved_label_len arg


Former-commit-id: 1771251ce3
2024-07-01 01:19:27 +08:00
hiyouga
654116c0b1 fix #4556
Former-commit-id: 59e0b4f616
2024-06-26 19:43:16 +08:00
hiyouga
d519c2fde5 tiny fix
Former-commit-id: 41086059b1
2024-06-25 01:15:19 +08:00
hoshi-hiyouga
709bbc1d92 Merge pull request #4417 from mMrBun/main
Add tool_format parameter to rewrite templates for different function call formats.

Former-commit-id: def6d280db
2024-06-24 23:17:55 +08:00
hoshi-hiyouga
b7f5cfde6e Update template.py
Former-commit-id: 1240bd57d8
2024-06-24 23:12:59 +08:00
hoshi-hiyouga
673f27a59e Update loader.py
Former-commit-id: dddfd516ee
2024-06-24 23:06:18 +08:00
hiyouga
47651a94a3 fix #4410
Former-commit-id: fca893d73c
2024-06-24 22:34:31 +08:00
mMrBun
c0e005e2ea Add tool_format to overwrite tool formatter template
Former-commit-id: 20e2e6fdcb
2024-06-22 02:13:23 +08:00
hiyouga
98abb5c900 remove dup template
Former-commit-id: db9a1912e3
2024-06-22 01:31:32 +08:00
hiyouga
3d72b1a856 fix jinja template
Former-commit-id: 2b596fb55f
2024-06-19 20:03:50 +08:00
hiyouga
7735456561 fix templates
Former-commit-id: 4cff6a4ad5
2024-06-19 17:44:05 +08:00
hiyouga
c9557241f6 fix bug
Former-commit-id: 6d2bf216ac
2024-06-19 03:49:23 +08:00
hiyouga
e73a235a38 use prefix to replace force system
Former-commit-id: 4f22eae8f4
2024-06-19 03:39:52 +08:00
hiyouga
bccc852f76 fix tool formatter, allow parallel function #4362
Former-commit-id: cd75b1fe9d
2024-06-19 03:23:51 +08:00
hoshi-hiyouga
6db02615d4 Merge pull request #4173 from mMrBun/main
Implemented the tool_formatter and tool_extractor for glm4 and Qwen2 tool_format

Former-commit-id: c0ca42566c
2024-06-19 03:18:55 +08:00
hiyouga
c0c6b8075a tiny fix
Former-commit-id: 38b6b0f52e
2024-06-16 01:06:41 +08:00
ancv
9d9f8c6531 remove some unused params
Former-commit-id: 04315c3d92
2024-06-15 23:00:55 +07:00
hiyouga
2946153cea add license
Former-commit-id: d87108daa6
2024-06-15 17:54:33 +08:00
hiyouga
8fccaf20c5 fix #4221
Former-commit-id: 6baafd4eb3
2024-06-13 02:48:21 +08:00
ancv
045eb155a2 implement efficient packing without cross-contamination attention
Former-commit-id: b2c367bc61
2024-06-12 11:56:01 +07:00
hoshi-hiyouga
bf3de9bfe8 Update pretrain.py
Former-commit-id: 0c29233237
2024-06-11 17:02:14 +08:00
d
da39715085 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
Former-commit-id: 6979f3f848
2024-06-11 16:23:40 +08:00
mMrBun
b6d63b3324 Optimize the handling of QWEN2 in scenarios involving multiple tool calls.
Former-commit-id: 950e360ca0
2024-06-10 02:00:14 +08:00
mMrBun
3f11ab800f Removed unnecessary comments.
Former-commit-id: 6ed0b0c800
2024-06-09 18:25:22 +08:00
mMrBun
daf472994d Merge branch 'hiyouga:main' into main
Former-commit-id: 0f2609ce19
2024-06-09 18:17:24 +08:00
mMrBun
18a86ea104 Implemented the tool_formatter and tool_extractor for glm4 tool_format
Former-commit-id: cb1cbcb293
2024-06-09 18:16:15 +08:00