Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / lucidrains/meshgpt-pytorch issues and pull requests

#96 - Recommended Hyperparamters

Issue - State: open - Opened by NickPJQ 23 days ago

#95 - does the “Int['b nv 3']” mean？

Issue - State: open - Opened by Syuao 24 days ago

#93 - hyper-parameter suggestion

Issue - State: closed - Opened by chinmay5 4 months ago - 20 comments

#92 - Area for quads

Issue - State: closed - Opened by sbriseid 5 months ago - 6 comments

#91 - Continuous range for discretize area

Issue - State: open - Opened by sbriseid 5 months ago - 2 comments

#90 - try to fix tests

Pull Request - State: closed - Opened by lucidrains 5 months ago

#89 - Version conflict with x-transformer-1.31.6

Issue - State: closed - Opened by hyeonjang 5 months ago - 1 comment

#88 - Main marcus main

Pull Request - State: closed - Opened by sbriseid 5 months ago - 1 comment

#87 - Fine gateloop doesnt use the param gateloop_use_heinsen

Pull Request - State: closed - Opened by MarcusLoppe 5 months ago - 19 comments

#86 - Added device property for autoencoder

Pull Request - State: closed - Opened by MarcusLoppe 5 months ago - 1 comment

#85 - try to figure out issue with gh actions

Pull Request - State: closed - Opened by lucidrains 5 months ago

#84 - Huggingface "from_pretrained" support

Pull Request - State: closed - Opened by MarcusLoppe 6 months ago

#83 - Area calculation for each triangle is same (with value of 0.5)

Issue - State: closed - Opened by aakash-saboo 6 months ago - 6 comments

#82 - Transformer keeps predicting the same token

Issue - State: closed - Opened by MagicianWu 6 months ago - 8 comments

#81 - Seeking Collaboration

Issue - State: closed - Opened by xhsyy 7 months ago - 3 comments

#79 - New/replace attention mechanism

Issue - State: closed - Opened by MarcusLoppe 7 months ago

#77 - Mesh conditioning instead of text conditioning

Issue - State: open - Opened by pathquester 7 months ago - 6 comments

#76 - Pretrained checkpoints?

Issue - State: open - Opened by agaier 8 months ago - 2 comments

#75 - High RAM cost in derive_face_edges_from_faces

Issue - State: closed - Opened by inmny 8 months ago

#73 - `ResidualLFQ` was successful, but `ResidualVQ` failed severely!

Issue - State: open - Opened by fighting-Zhang 8 months ago - 2 comments

#72 - A question about the code in 'meshgpt_pytorch.py'

Issue - State: open - Opened by LeeYan1 8 months ago

#71 - Mesh completion task

Issue - State: open - Opened by jrmgr 8 months ago - 2 comments

#70 - distirbution training

Issue - State: open - Opened by kingofrubbish2 8 months ago - 1 comment

#69 - Removed sageconv dropout due to not existing anymore(?)

Pull Request - State: closed - Opened by MarcusLoppe 9 months ago - 1 comment

#68 - Loss=nan when training transformertrainer

Issue - State: open - Opened by HenryLhc 9 months ago - 5 comments

#64 - GroupNorm in ResnetBlocks

Issue - State: closed - Opened by shanemankiw 9 months ago - 13 comments

#62 - Mesh intra face vertex id ordering convention

Issue - State: closed - Opened by jrmgr 9 months ago - 16 comments

#61 - Sliding window for transformer

Issue - State: open - Opened by MarcusLoppe 10 months ago - 1 comment

#60 - Just to get started

Issue - State: closed - Opened by studiohuahong 10 months ago - 2 comments

#59 - text_condition_model_types

Issue - State: closed - Opened by Ginjing-Yuan 10 months ago

#58 - Correction in the meshgpt paper.

Issue - State: closed - Opened by fire 10 months ago

#57 - Question

Issue - State: open - Opened by nicolasdonati 10 months ago - 10 comments

#56 - Is there a pretrained model and if not, how to train the model

Issue - State: closed - Opened by jonasweimar 10 months ago - 12 comments

#54 - Support for quads

Issue - State: closed - Opened by pathquester 10 months ago - 38 comments
Labels: enhancement

#53 - Isn't MeshGPT a VQ-VAE?

Issue - State: closed - Opened by rexxxx1234 10 months ago - 2 comments

#52 - is gaussian_blur_1d correct?

Issue - State: closed - Opened by thuliu-yt16 10 months ago - 1 comment

#49 - Training on my own data

Issue - State: closed - Opened by adeerAI 11 months ago - 1 comment

#48 - MeshTransformer.generate does not work with a prompt if kv cache is enabled

Issue - State: closed - Opened by Kurokabe 11 months ago - 15 comments

#47 - gateloop_use_heinsen=True on MeshTransformer results in NaN loss

Issue - State: closed - Opened by Kurokabe 11 months ago - 6 comments

#46 - Simple training script for toy data?

Issue - State: open - Opened by xiao-xian 11 months ago - 6 comments

#45 - Flash attention mispelling flash_attn -> attn_flash

Pull Request - State: closed - Opened by MarcusLoppe 11 months ago - 5 comments

#44 - Transformer - token_embed outputs nan values

Issue - State: closed - Opened by MarcusLoppe 11 months ago - 11 comments

#43 - Commit loss is negative

Issue - State: open - Opened by ZekaiGalaxy 11 months ago - 8 comments

#42 - Fix Shape Restoration of `face_mask` After `masked_fill` on `recon_faces`

Pull Request - State: closed - Opened by qixuema 11 months ago - 1 comment

#41 - transformer.generate()

Issue - State: closed - Opened by Genshin-Impact-king 11 months ago - 3 comments

#40 - `face_mask` rearrange in Autoencoder Forward Pass

Issue - State: closed - Opened by qixuema 11 months ago - 2 comments

#39 - Transformer - High VRAM, context length

Issue - State: open - Opened by MarcusLoppe 11 months ago - 4 comments

#38 - Recipe for overfitting a single obj

Issue - State: closed - Opened by ZekaiGalaxy 11 months ago - 6 comments

#37 - Proper dataset that reduces VRAM usage and provides higher performance.

Pull Request - State: open - Opened by MarcusLoppe 11 months ago - 14 comments

#36 - Residual quantization - VRAM bottleneck

Issue - State: closed - Opened by MarcusLoppe 11 months ago - 13 comments

#35 - How to preprocess ShapeNet models?

Issue - State: closed - Opened by xuanyaoming 11 months ago - 18 comments

#34 - derive_face_edges_from_faces high ram usage

Issue - State: closed - Opened by MarcusLoppe 11 months ago - 5 comments

#33 - Typo Error: `dim` Parameter in `MeshAutoencoder`

Issue - State: closed - Opened by qixuema 11 months ago - 1 comment

#32 - Commit Loss doesn't go down

Issue - State: closed - Opened by ZekaiGalaxy 11 months ago - 9 comments

#31 - use_heinsen

Issue - State: closed - Opened by Genshin-Impact-king 11 months ago - 2 comments

#30 - Handling EOS Token in Max Sequence Length Scenarios

Issue - State: closed - Opened by qixuema 11 months ago - 4 comments

#28 - Trainer, fails to load.

Issue - State: closed - Opened by MarcusLoppe 11 months ago - 7 comments

#27 - Avoid saving autoencoder checkpoint each step

Pull Request - State: closed - Opened by Kurokabe 11 months ago - 1 comment

#26 - Issue with Multiple `eos_token_id` in Code Sequences

Issue - State: closed - Opened by qixuema 11 months ago - 1 comment

#24 - Add pr for ArgumentParser CLI

Issue - State: open - Opened by fire 11 months ago

#23 - Brainstorm how to increase triangle counts by 10x

Issue - State: closed - Opened by fire 11 months ago - 1 comment

#22 - How to embed additional information like uvs / bones

Issue - State: closed - Opened by fire 11 months ago - 2 comments

#21 - Bad performance using local attention

Issue - State: closed - Opened by MarcusLoppe 11 months ago - 16 comments

#20 - Packed Sequences

Issue - State: closed - Opened by Algomancer 11 months ago - 3 comments

#19 - Can't run MeshTransformer.generate(return_codes = True)

Issue - State: closed - Opened by fire 11 months ago - 13 comments

#18 - The MeshTransformer does not generate coherent results

Issue - State: closed - Opened by Kurokabe 11 months ago - 15 comments

#17 - Mesh to Mesh training

Issue - State: closed - Opened by fire 11 months ago - 6 comments

#14 - custom_collate - only adding tensor to output

Issue - State: closed - Opened by MarcusLoppe 11 months ago - 1 comment

#13 - SageConv & ResNet sizes

Issue - State: closed - Opened by MarcusLoppe 11 months ago - 7 comments

#12 - Added train by epoch for Trainer and added support for texts

Pull Request - State: open - Opened by MarcusLoppe 12 months ago - 7 comments

#11 - Data augmentation strategies

Issue - State: open - Opened by fire 12 months ago - 56 comments

#10 - Mamba vs RWKV

Issue - State: closed - Opened by fire 12 months ago - 1 comment

#9 - Large memory requirement

Issue - State: closed - Opened by MarcusLoppe 12 months ago - 60 comments

#8 - How to evaluate the codes returning from the autoencoder?

Issue - State: closed - Opened by fire 12 months ago - 6 comments

#7 - How to view the generated mesh?

Issue - State: closed - Opened by Croolch 12 months ago - 2 comments

#6 - Add an example for converting gltf to inputs.

Pull Request - State: closed - Opened by fire 12 months ago - 48 comments

#4 - Post Process 3d Assets

Issue - State: open - Opened by adeerAI 12 months ago - 5 comments

#3 - Missing features for graph embedding

Issue - State: closed - Opened by MarcusLoppe 12 months ago - 20 comments

#2 - Use mesh autoencoder as is for chatml finetuning

Issue - State: closed - Opened by fire 12 months ago - 8 comments