FoundationVision/LlamaGen

Issues

• RuntimeError: shape '[1, 512, 1, 32, 2]' is invalid for input of size 16448

  #21 opened 5 months ago by beniz
  11

• About finetuning on pretrained LLM models.

  #73 opened 12 days ago by ZeyuLing
  0

• About `data_range` in ssim_loss

  #72 opened 14 days ago by xyfJASON
  0

• Why use such a big model---google/flan-t5-xl

  #71 opened 15 days ago by ranck626
  0

• 请问如何用我自己的数据集训练 Image tokenizers and AR models for text-conditional image generation？请问可以提供一个示例吗，谢谢

  #59 opened 3 months ago by gzhuinjune
  0

• KeyError: 'optimizer'

  #45 opened 4 months ago by sugary199
  4

• 你好，我使用您的代码训练了我自己的数据集，但是图片变得特别模糊

  #29 opened 5 months ago by gzhuinjune
  9

• dataloader的batch_size设置

  #70 opened 21 days ago by yyyxcleo
  0

• 和AIM的区别

  #69 opened 24 days ago by Jmh0527
  0

• StyleGAN vs PatchGAN

  #68 opened 24 days ago by sunset-clouds
  0

• About train losses and evalution parameters setting

  #56 opened 3 months ago by MrCrims
  1

• cfg-interval

  #67 opened a month ago by wangyf8848
  0

• Embbeding layer

  #66 opened a month ago by wangyf8848
  0

• How to reproduce the codebook usage

  #65 opened a month ago by BaohaoLiao
  0

• Why is the model GPT in the code？

  #64 opened a month ago by wangyf8848
  1

• Question about cannot reproduce FID results

  #53 opened 4 months ago by Ghy0501
  2

• The demo not work well

  #63 opened 2 months ago by bigbrother001
  0

• T2I performance on mscoco

  #47 opened 4 months ago by HalvesChen
  1

• The effect of VQVAE's training data on image generation

  #62 opened 2 months ago by HalvesChen
  0

• Recommendation for decoder finetuning

  #61 opened 2 months ago by elias-ramzi
  0

• Only inference

  #60 opened 2 months ago by heavenhellchen
  0

• When I set ipdb in gpt.py, I encounter this error,torch._dynamo.exc.InternalTorchDynamoError: `example_value` needs to be a `FakeTensor`wrapped by this instance of Dynamo. Found: tensor(..., device='meta', size=(2,))

  #58 opened 3 months ago by BinZhu-ece
  0

• About evaluation on private dataset

  #55 opened 4 months ago by MrCrims
  0

• About ROPE in sample process

  #54 opened 4 months ago by Leedonus
  6

• Loss increases during training the T2I model

  #51 opened 4 months ago by Epiphqny
  0

• Questions about the results of your experiment.

  #38 opened 5 months ago by potatowarriors
  2

• add t5 extraction instructions in Readme or Getting started for t2i training

  #52 opened 4 months ago by sahil02235
  0

• T2I VQVAE Training Details

  #50 opened 4 months ago by alexanderswerdlow
  0

• Mask guidance, inpaiting and outpaiting

  #49 opened 4 months ago by sahil02235
  7

• [Feature] ControlNet support via process similar to PixArt's ControlNet-Transformer

  #24 opened 5 months ago by kabachuha
  1

• Cannot Reproduce LlamaGen-B or L numbers using provided models

  #48 opened 4 months ago by vkramanuj
  1

• T2I Data

  #37 opened 4 months ago by HalvesChen
  0

• FID results of GPT-L and GPT-1B on 256*256 images

  #46 opened 4 months ago by LutingWang
  3

• FID Evaluation not matching paper results for VQ-16 checkpoint

  #34 opened 4 months ago by vkramanuj
  3

• Can LlamaGen predict a [EOS] token when inferencing?

  #44 opened 4 months ago by luminousking
  6

• Difficulty in reproducing results with pre-trained weights

  #41 opened 5 months ago by Rishit-dagli
  1

• Do you try class 2 Image generation with the image resolution of 512X512?

  #42 opened 5 months ago by OliverRensu
  0

• Training Results

  #39 opened 5 months ago by Huage001
  4

• tokenizer of 4 dim

  #40 opened 5 months ago by DidiD1
  0

• Questions about the discriminator

  #35 opened 5 months ago by Doctor-James
  1

• 你好，vq_ds16_c2i_training.pt 在私有的通用图片上finetune效果变差了，就试了在imgenet数据集接着finetune，效果也变差了，想问一下是哪里出问题了？

  #36 opened 5 months ago by 353xiong
  1

• Discriminator is not training properly?

  #27 opened 5 months ago by ThisisBillhe
  4

• Inquiry about the OpenImages dataset

  #32 opened 5 months ago by RobertLuo1
  2

• Text embedding inject

  #33 opened 5 months ago by daiyixiang666
  2

• Question about why not try using image tokenizer and a ready made llama3 etc LLM model with lora?

  #31 opened 5 months ago by lucasjinreal
  9

• Question about text-conditional generation.

  #30 opened 5 months ago by Yangr116
  2

• Which parameters are trainable? Are the encoder and decoder in VQGAN fixed? Is the llama fixed?

  #26 opened 5 months ago by tanshuai0219
  1

• Train script

  #28 opened 5 months ago by potatowarriors
  1

• [Feature] Inpainting script

  #25 opened 5 months ago by kabachuha
  0

• Mismatched model weights document

  #22 opened 5 months ago by Artanic30
  2