104 Commits

Author	SHA1	Message	Date
jack.butler	7036df8dc0	update link to relative link	2023-02-10 15:23:06 +00:00
jack.butler	3f9e2c31ac	add hyperlink to dummy yml file	2023-02-10 15:22:27 +00:00
jack.butler	2ebb336141	update readme	2023-02-10 15:18:31 +00:00
jack.butler	5de1c02a7a	add dummy yml config for reward	2023-02-10 15:18:03 +00:00
sanagnos	4dd0d67e9c	Merge pull request #1398 from jackapbutler/fix-tokenizer-match ... Add tests and update docstring to tokenizer matching	2023-02-10 11:37:00 +01:00
jack.butler	24b07523aa	add test for tokenizer matching behaviour	2023-02-10 09:47:20 +00:00
jack.butler	2fbf2fa457	add docstring info about tokenizer matching	2023-02-10 09:46:58 +00:00
jack.butler	090c5cbcc2	fix tokenizer matching and add tests	2023-02-09 18:47:38 +00:00
sanagnos	4ba622de8e	Merge branch 'main' into sft-data-sampling	2023-02-09 09:19:17 +01:00
Mark Worrall	9faae250ce	minor tidy-up	2023-02-08 20:57:13 +00:00
Mark Worrall	283df8ec84	Get working on multi-gpu	2023-02-08 20:49:25 +00:00
Mark Worrall	e2caf53654	First version of single GPU sampling working	2023-02-08 08:01:04 +00:00
theblackcat102	59dbfea48f	Merge pull request #1262 from jackapbutler/create-tokeniser-configs ... Add tokenizer config classes	2023-02-08 09:34:57 +08:00
hyunwoongko	cb722768f7	Apply pre-commit	2023-02-08 03:50:27 +09:00
hyunwoongko	44c555cad1	Add gelu fusion	2023-02-08 03:45:22 +09:00
jack.butler	eb1c4ada2a	rename arg name to model_name	2023-02-07 09:19:24 +00:00
jack.butler	dc7f255f01	add tokenizer config classes	2023-02-06 17:58:55 +00:00
Kian-Meng Ang	1e321a6fca	Fix typos (#1143) ... Found via `codespell -S .mypy_cache,yarn.lock,*.json,*.ipynb -L rouge,nam,vie`	2023-02-05 20:18:03 +01:00
sanagnos	222059b1b2	Merge pull request #991 from LAION-AI/rm-anthropic ... Add anthropic RLHF dataset & deepspeed support for reward model	2023-01-29 13:41:55 +01:00
theblackcat102	0e024e3955	[fix] Add working A100 config for deberta-xxlarge (deepspeed stuck during evaluation, deadlock?)	2023-01-29 03:17:11 +00:00
theblackcat102	fdcc629678	[feature] add reddit eli5, asks, askh; bug fix	2023-01-29 03:05:56 +00:00
theblackcat102	def03d75d2	[fix] Trim anthropic dataset down to last 2 convo only	2023-01-28 03:44:38 +00:00
theblackcat102	f43435efc9	[feature] add deepspeed default stage 2 config	2023-01-28 00:56:52 +00:00
theblackcat102	2a2f34391a	[fix] Added support for deepspeed	2023-01-28 00:55:40 +00:00
theblackcat102	3215a7bbf8	[feature] add initial version of anthropic dataset	2023-01-25 05:03:43 +00:00
theblackcat102	b8990d9078	[fix] remove spaces in format_pair	2023-01-23 02:48:47 +00:00
theblackcat102	736f46fb00	[fix] prosocial dialogue format error	2023-01-22 14:00:20 +00:00
theblackcat102	f5b2a34857	[feature] add pythia and limit translation pair	2023-01-22 00:56:17 +00:00
theblackcat102	62a203fd8c	[feature] move data formatting into dataset, instead of collator	2023-01-21 03:31:35 +00:00
theblackcat102	aca3e9de89	[fix] wait it pass?	2023-01-20 07:26:26 +00:00
theblackcat102	22e3ab1a89	[fix] linter fix	2023-01-20 07:23:02 +00:00
theblackcat102	c255148dc6	[fix] Merge main branch update	2023-01-20 06:18:43 +00:00
theblackcat102	6cd62e3d48	[fix] Fix missing russian and update readme	2023-01-20 06:16:26 +00:00
theblackcat102	74cb9aaa5a	[feature] added translation, rallio instruct tuning dataset, prosocial for safety, new summary dataset	2023-01-20 03:02:07 +00:00
Sotirios Anagnostidis	5d9591d82c	added soda dialogue dataset	2023-01-19 14:25:19 +01:00
theblackcat102	670be60ca8	[fix] Fix config typo	2023-01-14 12:17:58 +00:00
theblackcat102	6f6c590e57	[fix] Disable task specific evaluation	2023-01-14 06:47:21 +00:00
theblackcat102	1546111094	[feature] added GSM8k and code refactoring	2023-01-14 06:24:47 +00:00
theblackcat102	3966024871	[fix] Fix summarizer bug and QA typo issue	2023-01-14 05:49:22 +00:00
theblackcat102	9451aff6cc	[fix] @ekurtulus major logic bug in summarization	2023-01-14 03:49:19 +00:00
Sotirios Anagnostidis	c8f47eef9f	precommits	2023-01-11 22:58:17 +01:00
Sotirios Anagnostidis	d46ff8c4ee	better logging with deepspeed	2023-01-11 22:48:02 +01:00
Sotirios Anagnostidis	6438fdbe2c	quantization from #582	2023-01-11 22:44:20 +01:00
Sotirios Anagnostidis	4a3ea0b033	refactoring, now running	2023-01-11 22:42:04 +01:00
ekurtulus	5b77dd2e9f	better	2023-01-11 11:37:27 +03:00
mrcabbage972	d95c741ea0	Fixing requirements file	2023-01-10 20:16:02 -05:00
mrcabbage972	67aeed2cd7	Adding override of 32-bit optimization for embedding layer	2023-01-09 23:03:29 -05:00
mrcabbage972	08bdadf222	Adding BNB 8-bit Adam	2023-01-09 22:07:06 -05:00
theblackcat102	a1e1445de9	[fix] evaluation dataset is incorrect in reward-model/trainer.py	2023-01-08 03:42:40 +00:00
theblackcat102	9d05b73efc	[fix] syntax error and some typing issue in py38	2023-01-08 02:15:08 +00:00