Merge branch 'main' into audio_r2v

configs/attns/readme.md

-### TODO

configs/caching/adacache/wan_i2v_ada.json

+{
+    "infer_steps": 40,
+    "target_video_length": 81,
+    "target_height": 480,
+    "target_width": 832,
+    "self_attn_1_type": "flash_attn3",
+    "cross_attn_1_type": "flash_attn3",
+    "cross_attn_2_type": "flash_attn3",
+    "seed": 442,
+    "sample_guide_scale": 5,
+    "sample_shift": 3,
+    "enable_cfg": true,
+    "cpu_offload": false,
+    "feature_caching": "Ada"
+}

configs/caching/teacache/wan_t2v_14b.json → configs/changing_resolution/wan_t2v.json

@@ -12,15 +12,7 @@
     "sample_shift": 8,
     "enable_cfg": true,
     "cpu_offload": false,
-    "mm_config": {
-        "mm_type": "W-fp8-channel-sym-A-fp8-channel-sym-dynamic-Sgl",
-        "weight_auto_quant": true
-    },
-    "feature_caching": "Tea",
-    "coefficients": [
-        [-3.03318725e05, 4.90537029e04, -2.65530556e03, 5.87365115e01, -3.15583525e-01],
-        [-5784.54975374, 5449.50911966, -1811.16591783, 256.27178429, -13.02252404]
-    ],
-    "use_ret_steps": true,
-    "teacache_thresh": 0.26
+    "changing_resolution": true,
+    "resolution_rate": 0.75,
+    "changing_resolution_steps": 25
 }

configs/dist_infer/wan_i2v_dist_ring.json

+{
+    "infer_steps": 40,
+    "target_video_length": 81,
+    "target_height": 480,
+    "target_width": 832,
+    "self_attn_1_type": "flash_attn3",
+    "cross_attn_1_type": "flash_attn3",
+    "cross_attn_2_type": "flash_attn3",
+    "seed": 42,
+    "sample_guide_scale": 5,
+    "sample_shift": 5,
+    "enable_cfg": true,
+    "cpu_offload": false,
+    "parallel_attn_type": "ring",
+    "parallel_vae": true
+}

configs/dist_infer/wan_t2v_dist_ring.json

+{
+    "infer_steps": 50,
+    "target_video_length": 81,
+    "text_len": 512,
+    "target_height": 480,
+    "target_width": 832,
+    "self_attn_1_type": "flash_attn3",
+    "cross_attn_1_type": "flash_attn3",
+    "cross_attn_2_type": "flash_attn3",
+    "seed": 42,
+    "sample_guide_scale": 6,
+    "sample_shift": 8,
+    "enable_cfg": true,
+    "cpu_offload": false,
+    "parallel_attn_type": "ring",
+    "parallel_vae": true
+}

configs/distill/wan_i2v_distill_4step_cfg.json

@@ -11,5 +11,5 @@
     "sample_shift": 5,
     "enable_cfg": false,
     "cpu_offload": false,
-    "denoising_step_list": [999, 750, 500, 250]
+    "denoising_step_list": [1000, 750, 500, 250]
 }

configs/distill/wan_i2v_distill_4step_cfg_lora.json

 {
-    "infer_steps": 4,
-    "target_video_length": 81,
-    "target_height": 480,
-    "target_width": 832,
-    "self_attn_1_type": "flash_attn3",
-    "cross_attn_1_type": "flash_attn3",
-    "cross_attn_2_type": "flash_attn3",
-    "seed": 42,
-    "sample_guide_scale": 5,
-    "sample_shift": 5,
-    "enable_cfg": false,
-    "cpu_offload": false,
-    "denoising_step_list": [999, 750, 500, 250],
-    "lora_path": [
-        "Wan2.1-I2V-14B-480P/loras/Wan21_I2V_14B_lightx2v_cfg_step_distill_lora_rank64.safetensors"
-    ]
+  "infer_steps": 4,
+  "target_video_length": 81,
+  "target_height": 480,
+  "target_width": 832,
+  "self_attn_1_type": "flash_attn3",
+  "cross_attn_1_type": "flash_attn3",
+  "cross_attn_2_type": "flash_attn3",
+  "seed": 42,
+  "sample_guide_scale": 5,
+  "sample_shift": 5,
+  "enable_cfg": false,
+  "cpu_offload": false,
+  "denoising_step_list": [1000, 750, 500, 250],
+  "lora_configs": [
+    {
+      "path": "Wan2.1-I2V-14B-480P/loras/Wan21_I2V_14B_lightx2v_cfg_step_distill_lora_rank64.safetensors",
+      "strength": 1.0
+    }
+  ]
 }

configs/distill/wan_t2v_distill_4step_cfg.json

@@ -9,8 +9,8 @@
     "cross_attn_2_type": "flash_attn3",
     "seed": 42,
     "sample_guide_scale": 6,
-    "sample_shift": 8,
+    "sample_shift": 5,
     "enable_cfg": false,
     "cpu_offload": false,
-    "denoising_step_list": [999, 750, 500, 250]
+    "denoising_step_list": [1000, 750, 500, 250]
 }

configs/distill/wan_t2v_distill_4step_cfg_dynamic.json

@@ -9,10 +9,10 @@
     "cross_attn_2_type": "flash_attn3",
     "seed": 42,
     "sample_guide_scale": 6,
-    "sample_shift": 8,
+    "sample_shift": 5,
     "enable_cfg": false,
     "enable_dynamic_cfg": true,
     "cfg_scale": 4.0,
     "cpu_offload": false,
-    "denoising_step_list": [999, 750, 500, 250]
+    "denoising_step_list": [1000, 750, 500, 250]
 }

configs/distill/wan_t2v_distill_4step_cfg_lora.json

 {
-    "infer_steps": 4,
-    "target_video_length": 81,
-    "text_len": 512,
-    "target_height": 480,
-    "target_width": 832,
-    "self_attn_1_type": "flash_attn3",
-    "cross_attn_1_type": "flash_attn3",
-    "cross_attn_2_type": "flash_attn3",
-    "seed": 42,
-    "sample_guide_scale": 6,
-    "sample_shift": 8,
-    "enable_cfg": false,
-    "cpu_offload": false,
-    "denoising_step_list": [999, 750, 500, 250],
-    "lora_path": [
-        "Wan2.1-T2V-14B/loras/Wan21_T2V_14B_lightx2v_cfg_step_distill_lora_rank32.safetensors"
-    ]
+  "infer_steps": 4,
+  "target_video_length": 81,
+  "text_len": 512,
+  "target_height": 480,
+  "target_width": 832,
+  "self_attn_1_type": "flash_attn3",
+  "cross_attn_1_type": "flash_attn3",
+  "cross_attn_2_type": "flash_attn3",
+  "seed": 42,
+  "sample_guide_scale": 6,
+  "sample_shift": 5,
+  "enable_cfg": false,
+  "cpu_offload": false,
+  "denoising_step_list": [1000, 750, 500, 250],
+  "lora_configs": [
+    {
+      "path": "Wan2.1-T2V-14B/loras/Wan21_T2V_14B_lightx2v_cfg_step_distill_lora_rank64.safetensors",
+      "strength": 1.0
+    }
+  ]
 }

configs/offload/block/wan_i2v_block.json

@@ -11,9 +11,9 @@
     "sample_shift": 5,
     "enable_cfg": true,
     "cpu_offload": true,
+    "t5_cpu_offload": true,
     "offload_granularity": "block",
     "mm_config": {
-        "mm_type": "W-int8-channel-sym-A-int8-channel-sym-dynamic-Vllm",
-        "weight_auto_quant": true
+        "mm_type": "W-int8-channel-sym-A-int8-channel-sym-dynamic-Vllm"
     }
 }

configs/quantization/wan_i2v_quant_auto.json → configs/offload/block/wan_t2v_1_3b.json

 {
-    "infer_steps": 40,
+    "infer_steps": 4,
     "target_video_length": 81,
     "target_height": 480,
     "target_width": 832,
@@ -10,9 +10,11 @@
     "sample_guide_scale": 5,
     "sample_shift": 5,
     "enable_cfg": true,
-    "cpu_offload": false,
-    "mm_config": {
-        "mm_type": "W-int8-channel-sym-A-int8-channel-sym-dynamic-Vllm",
-        "weight_auto_quant": true
-    }
+    "t5_cpu_offload": true,
+    "t5_offload_granularity": "block",
+    "t5_quantized": true,
+    "t5_quantized_ckpt": "/path/to/models_t5_umt5-xxl-enc-fp8.pth",
+    "t5_quant_scheme": "fp8",
+    "unload_modules": true,
+    "use_tiling_vae": true
 }

configs/offload/block/wan_t2v_block.json

@@ -13,8 +13,8 @@
     "enable_cfg": true,
     "cpu_offload": true,
     "offload_granularity": "block",
+    "t5_cpu_offload": true,
     "mm_config": {
-        "mm_type": "W-int8-channel-sym-A-int8-channel-sym-dynamic-Vllm",
-        "weight_auto_quant": true
+        "mm_type": "W-int8-channel-sym-A-int8-channel-sym-dynamic-Vllm"
     }
 }

configs/offload/disk/wan_i2v_phase_lazy_load_480p.json

@@ -15,9 +15,9 @@
     "t5_offload_granularity": "block",
     "dit_quantized_ckpt": "/path/to/dit_quant_model",
     "mm_config": {
-        "mm_type": "W-fp8-channel-sym-A-fp8-channel-sym-dynamic-Vllm",
-        "weight_auto_quant": false
+        "mm_type": "W-fp8-channel-sym-A-fp8-channel-sym-dynamic-Vllm"
     },
+    "t5_cpu_offload": true,
     "t5_quantized": true,
     "t5_quantized_ckpt": "/path/to/models_t5_umt5-xxl-enc-fp8.pth",
     "t5_quant_scheme": "fp8",

configs/offload/disk/wan_i2v_phase_lazy_load_720p.json

@@ -15,9 +15,9 @@
     "t5_offload_granularity": "block",
     "dit_quantized_ckpt": "/path/to/dit_quant_model",
     "mm_config": {
-        "mm_type": "W-fp8-channel-sym-A-fp8-channel-sym-dynamic-Vllm",
-        "weight_auto_quant": false
+        "mm_type": "W-fp8-channel-sym-A-fp8-channel-sym-dynamic-Vllm"
     },
+    "t5_cpu_offload": true,
     "t5_quantized": true,
     "t5_quantized_ckpt": "/path/to/models_t5_umt5-xxl-enc-fp8.pth",
     "t5_quant_scheme": "fp8",