4188 Commits

This Branch

This Branch

All Branches

Author	SHA1	Message	Date
0cc4m	73d5ec0e5d	Pull latest gptq-changes	2023-03-29 20:07:26 +00:00
0cc4m	a0bc770426	Add basic groupsize support ... Write groupsize into filename, for example 4bit-128b.safetensors for groupsize 128	2023-03-29 19:49:05 +00:00
0cc4m	f6f7687cc0	Add 4bit safetensor support, improve loading code	2023-03-29 14:47:59 +00:00
0cc4m	8d008b87a6	Add OPT support	2023-03-29 13:27:11 +00:00
Digitous	e698f22706	Update README.md	2023-03-28 19:14:46 -04:00
0cc4m	ef6fe680a9	Fix high VRAM usage caused by workaround for scalar type error	2023-03-28 06:30:02 +00:00
henk717	66264d38c4	Add Mixes	2023-03-28 00:23:10 +02:00
0cc4m	0f1fc46078	Fix errors during inference	2023-03-27 21:30:43 +00:00
0cc4m	d1a2005a27	Add support for old and new 4-bit format. Old one needs 4bit-old.pt file to launch	2023-03-27 20:45:21 +00:00
Llama	157b1c75e7	Merge pull request #27 from henk717/united ... Merge united	2023-03-26 23:25:36 -07:00
0cc4m	2e7a8a1a66	Adapt KoboldAI to latest gptq changes	2023-03-27 04:48:21 +00:00
henk717	37c3fd00b9	Merge pull request #315 from jojorne/jojorne-patch-enable-renaming-deleting-wi-root-folder ... Enable renaming/deleting wi root folder by creating a new one	2023-03-26 16:30:41 +02:00
henk717	bbb554efd3	Merge branch 'KoboldAI:main' into united	2023-03-26 01:45:52 +01:00
0cc4m	9dcba38978	Pin transformers to a working Llama-compatible version	2023-03-24 19:07:28 +00:00
0cc4m	026eb3205e	Fix 4-bit loading error when not loading in 4-bit	2023-03-22 22:12:06 +00:00
0cc4m	8941428c66	Fix Kobold loading to CPU in 4-bit, causing CUDA ASSERT error	2023-03-22 06:22:34 +00:00
0cc4m	c7edc764b9	Fix llama loading	2023-03-21 21:58:31 +00:00
0cc4m	ecd065a881	Overhaul 4-bit support to load with a toggle	2023-03-21 21:40:59 +00:00
0cc4m	4cfc1219d4	Add gptq as submodule	2023-03-20 19:13:46 +00:00
0cc4m	3b7505dc28	Merge remote-tracking branch 'united/united' into 4bit	2023-03-20 19:06:40 +00:00
0cc4m	858657f669	Fix zipfile folder identification fix for Windows	2023-03-20 09:16:30 +01:00
somebody	91bb433b5f	GenericTokenizer: Fall back to defined tokenizer ... Shouldn't be relied on for model-agnostic code, but for loading processes where you know the tokenizer class used it should be okie dokie	2023-03-19 19:03:20 -05:00
0cc4m	60acf59316	Improve 4-bit llama support, add 4-bit gptj and gptneox support	2023-03-19 21:20:13 +00:00
henk717	94eb8ff825	TPU Message	2023-03-19 14:52:14 +01:00
Llama	5b8db52abb	Merge pull request #26 from henk717/united ... Merge united	2023-03-17 23:32:27 -07:00
somebody	864f9ed8c3	Add small debug for TPU	2023-03-17 17:06:48 -05:00
somebody	ffe85ce8a1	Modeling: Fix logits processors (probs, biasing, lua)	2023-03-17 16:56:47 -05:00
somebody	692dbfeb37	Merge branch 'united' of https://github.com/henk717/KoboldAI into model-structure-and-maybe-rwkv	2023-03-17 16:20:13 -05:00
somebody	8d0bc404a5	Model: More Jax import fixes and formatting	2023-03-17 15:36:44 -05:00
Henk	90a7eb6153	LLama tokenizer settings	2023-03-17 12:40:08 +01:00
Henk	1235b71bb5	Merge branch 'main' into united	2023-03-17 01:48:10 +01:00
Henk	219b824b9b	SocketIO Requirements Pin	2023-03-17 01:28:59 +01:00
henk717	86c87b23c0	Merge branch 'KoboldAI:main' into united	2023-03-17 01:17:58 +01:00
0cc4m	5d17692c79	Remove except Exception so that errors actually show up	2023-03-16 05:24:58 +00:00
YellowRoseCx	b3b454bbe4	Update huggingface.yml	2023-03-15 00:03:43 -05:00
YellowRoseCx	bf677a32f6	Merge remote-tracking branch 'catboxanon/test/4bit' into yr4bit	2023-03-14 17:09:06 -05:00
YellowRoseCx	2909910bcc	Merge branch 'henk717:united' into dev-yr	2023-03-14 17:04:33 -05:00
jojorne	8378ff9e26	Fix typo	2023-03-14 04:20:32 -03:00
jojorne	8d5a581d5d	Enable renaming deleting wi root folder by creating a new one	2023-03-14 04:04:01 -03:00
somebody	03af06638c	Modeling: Maybe fix samplers	2023-03-13 20:42:35 -05:00
somebody	65b60085e3	Undo debug	2023-03-13 20:30:46 -05:00
somebody	b93c339145	Model: Lazyload backends	2023-03-13 20:29:29 -05:00
somebody	adc11fdbc9	TPUMTJ: Fix loading bar ... I don't know why it works but I know it works	2023-03-13 20:13:05 -05:00
somebody	938c97b75a	RWKV: Fix yet another typo	2023-03-13 19:39:19 -05:00
somebody	3adc67c7a4	RWKV: Move import right before usage ... So we don't needlessly compile the cuda kernel	2023-03-13 19:37:45 -05:00
somebody	14b2543c7c	RWKV: Fix typo	2023-03-13 19:36:58 -05:00
somebody	b10b201701	Model: Add basic RWKV implementation	2023-03-13 19:34:38 -05:00
henk717	db7b53f52d	Merge pull request #310 from nkpz/united ... Fix out of range error after editing actions	2023-03-14 01:22:14 +01:00
catboxanon	5f3770bb58	Merge branch 'henk717:united' into test/4bit	2023-03-13 19:34:17 -04:00
somebody	bf8b60ac2d	Model: Add GenericTokenizer ... Because Hugging Face doesnt have a consistant API across their own libraries	2023-03-13 17:36:58 -05:00