Upload folder using huggingface_hub

diffusion/ckpts_new/decoder_step0100000.json

@@ -0,0 +1,18 @@
+{
+    "name": "SLatMeshDecodernew",
+    "args": {
+        "resolution": 64,
+        "model_channels": 768,
+        "phy_channels": 2048,
+        "latent_channels": 8,
+        "num_blocks": 12,
+        "num_heads": 12,
+        "mlp_ratio": 4,
+        "attn_mode": "swin",
+        "window_size": 8,
+        "use_fp16": true,
+        "representation_config": {
+            "use_color": true
+        }
+    }
+}

diffusion/ckpts_new/decoder_step0100000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a460d8ea685f97c3d320b03aabd3c5ae55cf94ef4e3bd1136860521f1075dcd
+size 1281392840

diffusion/ckpts_new/denoiser_phy_step0700000.json

@@ -0,0 +1,19 @@
+{
+    "name": "SLatFlowModelphy",
+    "args": {
+        "resolution": 64,
+        "in_channels": 8,
+        "out_channels": 8,
+        "model_channels": 1024,
+        "cond_channels": 1024,
+        "num_blocks": 14,
+        "num_heads": 16,
+        "mlp_ratio": 4,
+        "patch_size": 2,
+        "num_io_res_blocks": 2,
+        "io_block_channels": [128],
+        "pe_mode": "ape",
+        "qk_rms_norm": true,
+        "use_fp16": true
+    }
+}

diffusion/ckpts_new/denoiser_phy_step0700000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40e9b76b5f4c6663f87539151eb6b40b209e9bb2b09ef5f18feb58e8372a7a60
+size 1619037489

diffusion/ckpts_new/denoiser_step0700000.json

@@ -0,0 +1,21 @@
+{
+    "name": "SLatFlowModel",
+    "args": {
+        "resolution": 64,
+        "in_channels": 8,
+        "out_channels": 8,
+        "model_channels": 1024,
+        "cond_channels": 1024,
+        "num_blocks": 24,
+        "num_heads": 16,
+        "mlp_ratio": 4,
+        "patch_size": 2,
+        "num_io_res_blocks": 2,
+        "io_block_channels": [
+            128
+        ],
+        "pe_mode": "ape",
+        "qk_rms_norm": true,
+        "use_fp16": true
+    }
+}

diffusion/ckpts_new/denoiser_step0700000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abf3356a382e13e85e18c282a0963e5d3633b6bee1f29e974258c04a1ff1ac9b
+size 2822875445

diffusion/ckpts_new/property_decoder_step0100000.json

@@ -0,0 +1,17 @@
+{
+    "name": "PropertyDecoder",
+    "args": {
+        "resolution": 64,
+        "model_channels": 2048,
+        "latent_channels": 8,
+        "num_blocks": 4,
+        "num_heads": 16,
+        "mlp_ratio": 4,
+        "attn_mode": "swin",
+        "window_size": 8,
+        "use_fp16": true,
+        "representation_config": {
+            "use_color": true
+        }
+    }
+}

diffusion/ckpts_new/property_decoder_step0100000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab977e5c18afd73a60cee0a44a8c75bb93debabd041dbe1b914902f3dbab38d4
+size 1099504133

diffusion/ckpts_new/property_output_step0100000.json

@@ -0,0 +1,9 @@
+{
+    "name": "PropertyOutput",
+    "args": {
+        "model_channels": 32,
+        "output_channels_lang": 3072,
+        "output_channels_phy": 14,
+        "use_fp16": true
+    }
+}

diffusion/ckpts_new/property_output_step0100000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01ecaed44d9511e2b2dd800c4235e629567cba21053cd3b55cdb22008c32bbf2
+size 1099495424

diffusion/pipeline.json

@@ -0,0 +1,84 @@
+{
+    "name": "TrellisImageTo3DPipeline",
+    "args": {
+        "models": {
+            "sparse_structure_decoder": "/mnt/petrelfs/caoziang.p/3D_generation/phyX/docker_test/new/TRELLIS/pretrain/ckpts/ss_dec_conv3d_16l8_fp16",
+            "sparse_structure_flow_model": "/mnt/petrelfs/caoziang.p/3D_generation/phyX/docker_test/new/TRELLIS/pretrain/ckpts/ss_flow_img_dit_L_16l8_fp16",
+            "slat_decoder_gs": "/mnt/petrelfs/caoziang.p/3D_generation/phyX/docker_test/new/TRELLIS/pretrain/ckpts/slat_dec_gs_swin8_B_64l8gs32_fp16",
+            "slat_decoder_rf": "/mnt/petrelfs/caoziang.p/3D_generation/phyX/docker_test/new/TRELLIS/pretrain/ckpts/slat_dec_rf_swin8_B_64l8r16_fp16",
+
+            "slat_decoder_mesh": "ckpts_new/decoder_step0100000.pt",
+            "slat_decoder_phy": "ckpts_new/property_decoder_step0100000.pt",
+            "slat_decoder_output": "ckpts_new/property_output_step0100000.pt",
+            "slat_flow_model": "ckpts_new/denoiser_step0700000.pt",
+            "slat_flow_model_phy": "ckpts_new/denoiser_phy_step0700000.pt"
+        },
+        "sparse_structure_sampler": {
+            "name": "FlowEulerGuidanceIntervalSampler",
+            "args": {
+                "sigma_min": 1e-5
+            },
+            "params": {
+                "steps": 25,
+                "cfg_strength": 5.0,
+                "cfg_interval": [0.5, 1.0],
+                "rescale_t": 3.0
+            }
+        },
+        "slat_sampler": {
+            "name": "FlowEulerGuidanceIntervalSampler",
+            "args": {
+                "sigma_min": 1e-5
+            },
+            "params": {
+                "steps": 25,
+                "cfg_strength": 5.0,
+                "cfg_interval": [0.5, 1.0],
+                "rescale_t": 3.0
+            }
+        },
+        "slat_normalization": {
+            "mean": [
+                -2.1687545776367188,
+                -0.004347046371549368,
+                -0.13352349400520325,
+                -0.08418072760105133,
+                -0.5271206498146057,
+                0.7238689064979553,
+                -1.1414450407028198,
+                1.2039363384246826
+            ],
+            "std": [
+                2.377650737762451,
+                2.386378288269043,
+                2.124418020248413,
+                2.1748552322387695,
+                2.663944721221924,
+                2.371192216873169,
+                2.6217446327209473,
+                2.684523105621338
+            ],
+            "mean_phy": [
+                -2.1507165, 
+                -0.9456348, 
+                -2.0234883, 
+                -0.5949867, 
+                -3.608296 ,
+                -1.062877 , 
+                -3.288852 , 
+                -1.0749111
+            ],
+            "std_phy": [
+                0.6931998 , 
+                0.9221464 , 
+                0.6542199 , 
+                0.6594776 , 
+                0.8451334 ,
+                0.594917  , 
+                0.69759405, 
+                1.1614994 
+            ]
+        },
+        "image_cond_model": "dinov2_vitl14_reg"
+    }
+}

vae/ckpts/decoder_step0100000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a460d8ea685f97c3d320b03aabd3c5ae55cf94ef4e3bd1136860521f1075dcd
+size 1281392840

vae/ckpts/property_decoder_step0100000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab977e5c18afd73a60cee0a44a8c75bb93debabd041dbe1b914902f3dbab38d4
+size 1099504133

vae/ckpts/property_encoder_step0100000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:04e5ae16b65d2f29e014a898a06d7e555a3ad7f18a67ebe336f2f9208c860bcc
+size 1099500741

vae/ckpts/property_output_step0100000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01ecaed44d9511e2b2dd800c4235e629567cba21053cd3b55cdb22008c32bbf2
+size 1099495424

vae/config.json

@@ -0,0 +1,118 @@
+{
+    "models": {
+        "property_encoder": {
+            "name": "ElasticPropertyEncoder",
+            "args": {
+                "resolution": 64,
+                "in_channels": 3072,
+                "in_channels_phy": 14,
+                "model_channels": 768,
+                "latent_channels": 8,
+                "num_blocks": 4,
+                "num_heads": 12,
+                "mlp_ratio": 4,
+                "attn_mode": "swin",
+                "window_size": 8,
+                "use_fp16": true
+            }
+        },
+        "property_decoder": {
+            "name": "ElasticPropertyDecoder",
+            "args": {
+                "resolution": 64,
+                "model_channels": 2048,
+                "latent_channels": 8,
+                "num_blocks": 4,
+                "num_heads": 16,
+                "mlp_ratio": 4,
+                "attn_mode": "swin",
+                "window_size": 8,
+                "use_fp16": true,
+                "representation_config": {
+                    "use_color": true
+                }
+            }
+        },
+        "property_output": {
+            "name": "PropertyOutput",
+            "args": {
+                "model_channels": 32,
+                "output_channels_lang": 3072,
+                "output_channels_phy": 14,
+                "use_fp16": true
+            }
+        },
+        "decoder": {
+            "name": "ElasticSLatMeshDecodernew",
+            "args": {
+                "resolution": 64,
+                "model_channels": 768,
+                "phy_channels": 2048,
+                "latent_channels": 8,
+                "num_blocks": 12,
+                "num_heads": 12,
+                "mlp_ratio": 4,
+                "attn_mode": "swin",
+                "window_size": 8,
+                "use_fp16": true,
+                "representation_config": {
+                    "use_color": true
+                }
+            }
+        }
+    },
+    "dataset": {
+        "name": "Slat2RenderGeomesh",
+        "args": {
+            "image_size": 384,
+            "latent_model": "dinov2_vitl14_reg_slat_enc_swin8_B_64l8_fp16",
+            "min_aesthetic_score": 4.5,
+            "max_num_voxels": 28000
+        }
+    },
+    "trainer": {
+        "name": "SLatVaeMeshTrainer",
+        "args": {
+            "onlyphy_property": true,
+            "max_steps": 1000000,
+            "batch_size_per_gpu": 4,
+            "batch_split": 4,
+            "optimizer": {
+                "name": "AdamW",
+                "args": {
+                    "lr": 0.0001,
+                    "weight_decay": 0.0
+                }
+            },
+            "ema_rate": [
+                0.9999
+            ],
+            "fp16_mode": "inflat_all",
+            "fp16_scale_growth": 0.001,
+            "elastic": {
+                "name": "LinearMemoryController",
+                "args": {
+                    "target_ratio": 0.6,
+                    "max_mem_ratio_start": 0.5
+                }
+            },
+            "grad_clip": {
+                "name": "AdaptiveGradClipper",
+                "args": {
+                    "max_norm": 1.0,
+                    "clip_percentile": 95
+                }
+            },
+            "i_log": 10,
+            "i_sample": 5000,
+            "i_save": 10000,
+            "lambda_ssim": 0.2,
+            "lambda_lpips": 0.2,
+            "lambda_tsdf": 0.01,
+            "lambda_depth": 10.0,
+            "lambda_color": 0.1,
+            "lambda_kl": 1e-06,
+            "depth_loss_type": "smooth_l1"
+        }
+    }
+}