update configs for causvid (#34)

configs/wan_t2v_causvid.json

@@ -8,11 +8,7 @@
     "sample_guide_scale": 6,
     "sample_shift": 8,
     "enable_cfg": false,
-    "cpu_offload": true,
-    "mm_config": {
-        "mm_type": "W-fp8-channel-sym-A-fp8-channel-sym-dynamic-Sgl",
-        "weight_auto_quant": true
-    },
+    "cpu_offload": false,
     "num_fragments": 3,
     "num_frames": 21,
     "num_frame_per_block": 3,

configs/wan_t2v_causvid_save_quant.json

+{
+    "infer_steps": 9,
+    "target_video_length": 81,
+    "target_height": 480,
+    "target_width": 832,
+    "attention_type": "flash_attn3",
+    "seed": 42,
+    "sample_guide_scale": 6,
+    "sample_shift": 8,
+    "enable_cfg": false,
+    "cpu_offload": false,
+    "mm_config": {
+        "mm_type": "W-fp8-channel-sym-A-fp8-channel-sym-dynamic-Sgl"
+    },
+    "num_fragments": 3,
+    "num_frames": 21,
+    "num_frame_per_block": 3,
+    "num_blocks": 7,
+    "frame_seq_length": 1560,
+    "denoising_step_list": [999, 934, 862, 756, 603, 410, 250, 140, 74]
+}

scripts/run_wan_t2v_causvid.sh

@@ -35,4 +35,4 @@ python -m lightx2v.infer \
 --config_json ${lightx2v_path}/configs/wan_t2v_causvid.json \
 --prompt "Two anthropomorphic cats fight intensely on a spotlighted stage; the left cat wearing blue boxing gear with matching gloves, the right cat in bright red boxing attire and gloves." \
 --negative_prompt 色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走 \
---save_video_path ${lightx2v_path}/save_results/output_lightx2v_wan_t2v_causal.mp4
+--save_video_path ${lightx2v_path}/save_results/output_lightx2v_wan_t2v_causvid.mp4

scripts/run_wan_t2v_causvid_save_quant.sh

+#!/bin/bash
+
+# set path and first
+lightx2v_path=
+model_path=
+
+# check section
+if [ -z "${CUDA_VISIBLE_DEVICES}" ]; then
+    cuda_devices=0
+    echo "Warn: CUDA_VISIBLE_DEVICES is not set, using default value: ${cuda_devices}, change at shell script or set env variable."
+    export CUDA_VISIBLE_DEVICES=${cuda_devices}
+fi
+
+if [ -z "${lightx2v_path}" ]; then
+    echo "Error: lightx2v_path is not set. Please set this variable first."
+    exit 1
+fi
+
+if [ -z "${model_path}" ]; then
+    echo "Error: model_path is not set. Please set this variable first."
+    exit 1
+fi
+
+export TOKENIZERS_PARALLELISM=false
+
+export PYTHONPATH=${lightx2v_path}:$PYTHONPATH
+
+export ENABLE_PROFILING_DEBUG=true
+export ENABLE_GRAPH_MODE=false
+
+
+# =========================
+# save quantization weight
+# =========================
+
+export RUNNING_FLAG=save_naive_quant
+
+python -m lightx2v.infer \
+--model_cls wan2.1_causvid \
+--task t2v \
+--model_path $model_path \
+--config_json ${lightx2v_path}/configs/wan_t2v_causvid_save_quant.json \
+--prompt "Two anthropomorphic cats in comfy boxing gear and bright gloves fight intensely on a spotlighted stage." \
+--negative_prompt 色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走 \
+--save_video_path ${lightx2v_path}/save_results/output_lightx2v_wan_t2v_causvid.mp4
+
+sleep 2
+
+# =========================
+# load quantization weight and inference
+# =========================
+
+export RUNNING_FLAG=infer
+
+python -m lightx2v.infer \
+--model_cls wan2.1_causvid \
+--task t2v \
+--model_path $model_path \
+--config_json ${lightx2v_path}/configs/wan_t2v_causvid_save_quant.json \
+--prompt "Two anthropomorphic cats in comfy boxing gear and bright gloves fight intensely on a spotlighted stage." \
+--negative_prompt 色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走 \
+--save_video_path ${lightx2v_path}/save_results/output_lightx2v_wan_t2v_causvid.mp4