130 Commits

Author SHA1 Message Date
hiyouga
9a3b4d9ee0 fix lora target
Former-commit-id: d822e41e7ac7e310ee49e347fc45754284ce30b8
2023-09-09 17:04:45 +08:00
hiyouga
34da72ffbb support lora target auto find
Former-commit-id: bce9984733d88bf013847eed523d1c75fdf0995e
2023-09-09 15:38:37 +08:00
hiyouga
b4e2433e72 fix chatglm2 tokenizer
Former-commit-id: 1ab60b4a93fa1be5dfe6ffbd4deb64c0f9d9b431
2023-09-09 13:50:29 +08:00
hiyouga
9f52ff8cd6 fix bug in DPO data collator
Former-commit-id: 4fc262cdf1347691e253bdfbd96568db5a49c086
2023-09-08 20:45:07 +08:00
hiyouga
19dc1dbc64 fix #761
Former-commit-id: be76f6cbe5143f781b6b39603b80392253b3080a
2023-09-08 20:22:18 +08:00
hiyouga
801d1fa7b9 change to right-padding, update reward score #803
Former-commit-id: baa90415bc8f5ebd423d001378b51c3a3a6c2ec7
2023-09-08 20:04:31 +08:00
hiyouga
8b7eaee38d update requirements
Former-commit-id: d796a4a5709c390629bafbeb7c91fccf6a9076d0
2023-09-07 19:26:25 +08:00
hiyouga
970fd3944b fix #818
Former-commit-id: e81fd458c279ed2f3cee780e517482b425c8886d
2023-09-07 19:19:53 +08:00
hiyouga
52871b35b5 add deepspeed check in PPO training
Former-commit-id: e203ec7f71f504ccbaa89c27d20b8a0d9fa53f7e
2023-09-07 19:12:40 +08:00
hiyouga
c426dd6d69 fix #809
Former-commit-id: 2783ca75365d7c373cefba039788a48f0b8f35fc
2023-09-07 19:04:32 +08:00
hiyouga
47d9325873 fix baichuan templates
Former-commit-id: f48a49e835b32f3991cfad8874c7b9c78953809f
2023-09-07 18:54:14 +08:00
hiyouga
c5fcf5b3a5 refactor dataset_attr, add eos in pt, fix #757
Former-commit-id: 0feec9a830b917b36686b61938a66e842eccf930
2023-09-01 19:00:45 +08:00
codemayq
accf14aa9a add dataset stage check
Former-commit-id: 5c719a7ce988339d034a653456da9742dc2cec7c
2023-08-30 16:23:08 +08:00
hiyouga
618e3ab83c fix import error
Former-commit-id: b3207a974a45038591b8cbbcf20d1ca1142d6679
2023-08-23 20:45:03 +08:00
hiyouga
67c0d56bd9 fix #649
Former-commit-id: e6120a937ddb4f3c0b9bcb2466742f5cf4f77f8c
2023-08-23 20:21:15 +08:00
hiyouga
8b1ea24129 fix #617
Former-commit-id: a7bdaf1c92c7d798caf8438dc42a8972632ec584
2023-08-21 18:16:11 +08:00
hiyouga
922ca89c1d fix ppo trainer #551
Former-commit-id: 050a5447c191b8c50a0826a0f03bae499bff8b48
2023-08-20 14:07:11 +08:00
hiyouga
85abbe1f40 Release v0.1.7
Former-commit-id: 81abe8d6cabaa1ebe74dc32a5dc143389e4c9f31
2023-08-18 17:21:27 +08:00
hiyouga
8c841d6f76 tiny fix
Former-commit-id: 0ee159654ac6339c162745b004e2152ba6fe3c81
2023-08-18 13:07:35 +08:00
hiyouga
68c30094e1 support ppo score norm (trl 0.5.1.dev required)
Former-commit-id: 2b25db6d260ec1532281a592e873579346c7d21c
2023-08-18 12:02:42 +08:00
hiyouga
a8dd39ad08 fix PPO trainer #551 , update readme
Former-commit-id: faead74849470cebae9e37cde5fab2a71b32aa43
2023-08-18 11:43:10 +08:00
hiyouga
c0e887c85a update training resuming
Former-commit-id: 2ec75c31f609e65116ac3b621eeb7d8ccbf69135
2023-08-18 01:41:17 +08:00
hoshi-hiyouga
c856724d3f Merge branch 'main' into main
Former-commit-id: 870d2c7bf74d0da5a927bef4b8b01d15cc66a3e9
2023-08-18 01:37:23 +08:00
hiyouga
ff5ebb6a86 support bf16 ppo #551
Former-commit-id: 092088967de7409a2d51847cfc7afc83a8887320
2023-08-18 00:40:32 +08:00
hiyouga
a089d2665a fix ChatGLM2 ppo #527 #528
Former-commit-id: 60d6ad64d7c9f6445b0df8de0153c3a311974198
2023-08-18 00:34:59 +08:00
hiyouga
bb7028f7e2 fix generation bug #532
Former-commit-id: c071121e67374e5f09798db57cfc8668617a36ae
2023-08-17 22:21:34 +08:00
hiyouga
4319b5172c fix streaming in pt stage #548 #549
Former-commit-id: 050e992bee2a9293cc7399b578de807b5bf9bddc
2023-08-17 17:59:26 +08:00
hiyouga
22dec02b5f fix generation
Former-commit-id: 66a0300d312ef91c24fcf80667fa3b0bb8e1a342
2023-08-16 22:39:54 +08:00
hiyouga
1449d3182c fix system prompt
Former-commit-id: 411e775aa939bdd154a3f1e92921ede90d989f18
2023-08-16 01:35:52 +08:00
hiyouga
0684a5adc8 fix ChatGLM RLHF
Former-commit-id: 4e43e887e432ceb7e9287b4e309b63af3c3ba1bf
2023-08-15 11:19:20 +08:00
hiyouga
4d6534a665 alert pad_token source
Former-commit-id: f26a84e0d927d2554890daf431a93652e18f4235
2023-08-15 00:07:56 +08:00
hiyouga
1ce4a6109f fix ChatGLM lm_head #494
Former-commit-id: bf0048abdaeb2b9592d38ac991704ad014370b47
2023-08-14 14:14:48 +08:00
hiyouga
fff747ee5a web UI integrating RLHF
Former-commit-id: 137fd146b90f89a1164b56e6d507b30b1f5c2437
2023-08-14 10:48:47 +08:00
hiyouga
0366115d29 fix #480
Former-commit-id: ec15ca8fffacba2c34e1849c5ce90ca9989d66a2
2023-08-14 00:23:56 +08:00
hiyouga
03537dda06 tiny fix
Former-commit-id: 50a34c043de6d9e1410291e1d8c1ea9d53754e9e
2023-08-12 22:02:43 +08:00
hiyouga
4bbbee4370 fix rope scaling
Former-commit-id: 2e0dd36700ec5e8294581c1db4b9431f755fc5f8
2023-08-12 22:00:01 +08:00
hiyouga
56d6448c4a update readme
Former-commit-id: 6fa381400c21fa249cebcdff8c3afd72f8de20b3
2023-08-12 21:00:11 +08:00
hiyouga
a15a602809 support rope scaling, fix #475 #476 #478
Former-commit-id: 337d5f68b72230e545e7a94ca789187c7a2b7187
2023-08-12 20:46:27 +08:00
hiyouga
9644286027 fix unusual output of 8bit models #278 #391
Former-commit-id: 337ce5272b81f5561162beb08814b0e5abf23703
2023-08-12 00:25:29 +08:00
hiyouga
98aa629843 Release v0.1.6
Former-commit-id: 43c8b3c3c8bfb2e32d17fb3e8b194938e37d54bd
2023-08-11 23:25:57 +08:00
hiyouga
7ada4f5f6f support DPO training (2305.18290)
Former-commit-id: 6d98de148e4af63a7028dfaeb6cf86eb56a4488f
2023-08-11 03:02:53 +08:00
hiyouga
6eb4120464 support val set in streaming mode
Former-commit-id: faed15b58ed00b1e09bb091e7eee48f5ef7c508b
2023-08-09 23:00:26 +08:00
niuba
186d7f2d67 add last_checkpoint support
Former-commit-id: 9f1977e4de00b14a9d1b555c25bcaf12998d5046
2023-08-09 16:39:27 +08:00
hiyouga
3a821e5f0c fix sft trainer
Former-commit-id: 08cc888b1569572d0cd20bcf3f07e20072a0311a
2023-08-09 16:35:03 +08:00
hiyouga
b6316b460f fix tokenizer #417
Former-commit-id: 01aa678311bfd213a4b410a4e0ff09f48a0d40a1
2023-08-08 23:59:41 +08:00
hiyouga
5ada0ee44c fix bug
Former-commit-id: c13ce66021b21e015871b84489eeafa127a424a4
2023-08-08 17:55:55 +08:00
hiyouga
ab66cc8e37 fix chatml template #408
Former-commit-id: 21e0cc3f44c35ae689b00b274391492f413725ac
2023-08-08 17:44:39 +08:00
hiyouga
1ede9595c1 fix #376
Former-commit-id: a5b01257ba3323bcb2dd0217fb89a387e39ddbec
2023-08-07 13:58:59 +08:00
hiyouga
1d0ccf63f2 update trainer
Former-commit-id: 0d39b53a5164e34d22fe0a492eaa0d7ac63102fe
2023-08-07 13:34:35 +08:00
hiyouga
577b3cd0de fix qwen eos token
Former-commit-id: 770830c67886f5872b39b9608949ec62d4616b27
2023-08-06 13:31:17 +08:00