590 Commits

This Branch

This Branch

All Branches

Author	SHA1	Message	Date
Phil Wang	848e8a480a	always rederive the predicted noise from the clipped x0 for ddim + predict noise objective 1.12.3	2023-03-05 10:45:44 -08:00
Phil Wang	cc58f75474	bump to newer package of clip-anytorch that allows for text encodings < maximum context length 1.12.2	2023-03-04 09:37:25 -08:00
Phil Wang	3b2cf7b0bc	fix for self conditioning in diffusion prior network https://github.com/lucidrains/DALLE2-pytorch/issues/273 1.12.1	2023-02-11 17:18:40 -08:00
Phil Wang	984d62a373	default ddim sampling eta to 0 1.12.0	2022-12-23 13:23:09 -08:00
Phil Wang	683dd98b96	extra insurance in case eos id is not there 1.11.4	2022-12-15 10:54:21 -08:00
Phil Wang	067ac323da	address https://github.com/lucidrains/DALLE2-pytorch/issues/266 1.11.2	2022-11-23 08:41:25 -08:00
zion	91c8d1ca13	bug fix cosine annealing optimizer in prior trainer (#262)	2022-11-11 12:15:13 -08:00
zion	08238a7200	depend on open-clip-torch (#261) ... fix the previous commit which assumes open_clip is installed	2022-11-07 16:19:08 -08:00
zion	7166ad6711	add open clip to train_config (#260) ... add the ability to use open_clip in the train configs (useful for the new SOTA h/14 model)	2022-11-07 15:44:36 -08:00
Phil Wang	fbba0f9aaf	bring in prediction of v objective, combining the findings from progressive distillation paper and imagen-video to the eventual extension of dalle2 to make-a-video 1.11.1	2022-10-28 18:21:07 -07:00
Romain Beaumont	9f37705d87	Add static graph param (#226) ... * Add static graph param * use static graph param	2022-10-25 19:31:29 +02:00
Phil Wang	c3df46e374	fix openclipadapter to be able to use latest open sourced sota model 1.10.9	2022-10-23 15:12:09 -07:00
Phil Wang	41fabf2922	fix a dtype conversion issue for the diffusion timesteps in the diffusion prior, thanks to @JiaHeng-DLUT 0.10.8	2022-10-19 09:26:06 -07:00
Heng Jia	5975e8222b	Fix assert message (#253)	2022-10-18 08:50:59 -07:00
Phil Wang	c18c080128	fix for use with larger openai clip models by extracting dimension of last layernorm in clip 1.10.7	2022-09-29 09:09:47 -07:00
Phil Wang	b39653cf96	fix readme dataloader example	2022-09-20 08:39:52 -07:00
Phil Wang	39f8b6cf16	show example of using SOTA open sourced open clip	2022-09-19 10:45:20 -07:00
Phil Wang	d0c11b30b0	handle open clip adapter image size being a tuple 1.10.6	2022-09-19 10:27:14 -07:00
zion	86e2d5ba84	Minor Decoder Train Script Fixes (#242) ... * ensure tokenized text is on proper device * fix lpips mage distribution	2022-09-15 17:21:48 -07:00
Phil Wang	0d82dff9c5	in ddim, noise should be predicted after x0 is maybe clipped, thanks to @lukovnikov for pointing this out in another repository 1.10.5	2022-09-01 09:40:47 -07:00
Phil Wang	8bbc956ff1	fix bug with misnamed variable in diffusion prior network 1.10.4	2022-08-31 17:19:05 -07:00
Phil Wang	22019fddeb	todo	2022-08-31 13:36:05 -07:00
Phil Wang	6fb7e91343	fix ddim to use alpha_cumprod 1.10.3	2022-08-31 07:40:46 -07:00
Phil Wang	ba58ae0bf2	add two asserts to diffusion prior to ensure matching image embedding dimensions for clip, diffusion prior network, and what was set on diffusion prior 1.10.1	2022-08-28 10:11:37 -07:00
Phil Wang	1cc5d0afa7	upgrade to best downsample 1.10.0	2022-08-25 10:37:02 -07:00
Phil Wang	59fa101c4d	fix classifier free guidance for diffusion prior, thanks to @jaykim9870 for spotting the issue 1.9.0	2022-08-23 08:29:01 -07:00
Aidan Dempster	916ece164c	Merge pull request #234 from Veldrovive/deepspeed_fp16 ... Fixed issues with clip and deepspeed fp16	2022-08-20 19:01:43 -04:00
Aidan	cbaadb6931	Fixed issues with clip and deepspeed fp16 ... Also more more general compatibility fixes	2022-08-20 17:58:32 +00:00
Phil Wang	083508ff8e	cast attention matrix back to original dtype pre-softmax in attention 1.8.4	2022-08-20 10:56:01 -07:00
Phil Wang	7762edd0ff	make it work for @ethancohen123	2022-08-19 11:28:58 -07:00
Phil Wang	de5e628773	cite einops	2022-08-17 08:58:41 -07:00
Phil Wang	1b4046b039	gratitude	2022-08-17 08:57:33 -07:00
Phil Wang	27f19ba7fa	make sure diffusion prior trainer can operate with no warmup 1.8.2	2022-08-15 14:27:40 -07:00
Phil Wang	8f38339c2b	give diffusion prior trainer cosine annealing lr too 1.8.1	2022-08-15 07:38:01 -07:00
Phil Wang	6b9b4b9e5e	add cosine annealing lr schedule 1.8.0	2022-08-15 07:29:56 -07:00
Phil Wang	44e09d5a4d	add weight standardization behind feature flag, which may potentially work well with group norm	2022-08-14 11:34:45 -07:00
Phil Wang	34806663e3	make it so diffusion prior p_sample_loop returns unnormalized image embeddings 1.6.5	2022-08-13 10:03:40 -07:00
Phil Wang	dc816b1b6e	dry up some code around handling unet outputs with learned variance 1.6.4	2022-08-12 15:25:03 -07:00
Phil Wang	05192ffac4	fix self conditioning shape in diffusion prior 1.6.3	2022-08-12 12:30:03 -07:00
Phil Wang	9440411954	make self conditioning technique work with diffusion prior 1.6.1	2022-08-12 12:20:51 -07:00
Phil Wang	981d407792	comment	2022-08-12 11:41:23 -07:00
Phil Wang	7c5477b26d	bet on the new self-conditioning technique out of geoffrey hintons group 1.6.0	2022-08-12 11:36:08 -07:00
Phil Wang	be3bb868bf	add gradient checkpointing for all resnet blocks 1.5.0	2022-08-02 19:21:44 -07:00
Phil Wang	451de34871	enforce clip anytorch version 1.4.6	2022-07-30 10:07:55 -07:00
Phil Wang	f22e8c8741	make open clip available for use with dalle2 pytorch 1.4.5	2022-07-30 09:02:31 -07:00
Phil Wang	87432e93ad	quick fix for linear attention 1.4.4	2022-07-29 13:17:12 -07:00
Phil Wang	d167378401	add cosine sim for self attention as well, as a setting 1.4.3	2022-07-29 12:48:20 -07:00
Phil Wang	2d67d5821e	change up epsilon in layernorm the case of using fp16, thanks to @Veldrovive for figuring out this stabilizes training 1.4.2	2022-07-29 12:41:02 -07:00
Phil Wang	748c7fe7af	allow for cosine sim cross attention, modify linear attention in attempt to resolve issue on fp16 1.4.0	2022-07-29 11:12:18 -07:00
Phil Wang	80046334ad	make sure entire readme runs without errors 1.2.2	2022-07-28 10:17:43 -07:00