Upload 2 files

inference_512_v1.0_09B.yaml

@@ -0,0 +1,105 @@
+model:
+  target: lvdm.models.ddpm3d.LatentVisualDiffusion
+  params:
+          #rescale_betas_zero_snr: True
+    rescale_betas_zero_snr: False
+    parameterization: "v"
+    #parameterization: "eps"
+    linear_start: 0.00085
+    linear_end: 0.012
+    num_timesteps_cond: 1
+    timesteps: 1000
+    first_stage_key: video
+    cond_stage_key: caption
+    cond_stage_trainable: False
+    conditioning_key: hybrid
+    image_size: [40, 64]
+    channels: 4
+    scale_by_std: False
+    scale_factor: 0.18215
+    use_ema: False
+    uncond_type: 'empty_seq'
+    use_dynamic_rescale: true
+    base_scale: 0.7
+    fps_condition_type: 'fps'
+    perframe_ae: True
+    unet_config:
+      target: lvdm.modules.networks.openaimodel3d.UNetModel
+      params:
+        in_channels: 8
+        out_channels: 4
+        model_channels: 320
+        attention_resolutions:
+        - 4
+        - 2
+        - 1
+        num_res_blocks: 1
+        channel_mult:
+        - 1
+        - 2
+        - 4
+        - 4
+        dropout: 0.1
+        num_head_channels: 64
+        transformer_depth: 1
+        context_dim: 1024
+        use_linear: true
+        use_checkpoint: True
+        temporal_conv: True
+        temporal_attention: True
+        temporal_selfatt_only: true
+        use_relative_position: false
+        use_causal_attention: False
+        temporal_length: 16
+        addition_attention: true
+        image_cross_attention: true
+        default_fs: 24
+        fs_condition: true
+
+    first_stage_config:
+      target: lvdm.models.autoencoder.AutoencoderKL
+      params:
+        embed_dim: 4
+        monitor: val/rec_loss
+        ddconfig:
+          double_z: True
+          z_channels: 4
+          resolution: 256
+          in_channels: 3
+          out_ch: 3
+          ch: 128
+          ch_mult:
+          - 1
+          - 2
+          - 4
+          - 4
+          num_res_blocks: 2
+          attn_resolutions: []
+          dropout: 0.0
+        lossconfig:
+          target: torch.nn.Identity
+
+    cond_stage_config:
+      target: lvdm.modules.encoders.condition.FrozenOpenCLIPEmbedder
+      params:
+        freeze: true
+        layer: "penultimate"
+
+    img_cond_stage_config:
+      target: lvdm.modules.encoders.condition.FrozenOpenCLIPImageEmbedderV2
+      params:
+        freeze: true
+    
+    image_proj_stage_config:
+      target: lvdm.modules.encoders.resampler.Resampler
+      params:
+        dim: 1024
+        depth: 4
+        dim_head: 64
+        heads: 12
+        num_queries: 16
+        embedding_dim: 1280
+        output_dim: 1024
+        ff_mult: 4
+        video_length: 16
+

run_hummingbird.sh

@@ -0,0 +1,51 @@
+version=$1  ## 1024, 512, 256
+GPU=3
+ckpt='./stage_1.ckpt'
+config='configs/inference_512_v1.0_09B.yaml'
+prompt_dir="/VBench"
+base_res_dir="result"
+
+# 判断分辨率
+if [ "$1" == "256" ]; then
+    H=256
+    FS=3
+elif [ "$1" == "512" ]; then
+    H=320
+    FS=24
+elif [ "$1" == "1024" ]; then
+    H=576
+    FS=10
+else
+    echo "Invalid input. Please enter 256, 512, or 1024."
+    exit 1
+fi
+
+
+seed='123'
+sub_dir='0'
+res_dir="${base_res_dir}/${sub_dir}"
+
+echo "Running seed=$seed -> Saving to: $res_dir"
+echo $prompt_dir
+
+CUDA_VISIBLE_DEVICES=$GPU python3 scripts/evaluation/inference.py \
+    --seed ${seed} \
+    --ckpt_path $ckpt \
+    --config $config \
+    --savedir $res_dir \
+    --n_samples 1 \
+    --bs 1 --height ${H} --width $1 \
+    --unconditional_guidance_scale 7.5 \
+    --ddim_steps 16 \
+    --ddim_eta 1.0 \
+    --prompt_dir $prompt_dir \
+    --text_input \
+    --video_length 16 \
+    --frame_stride ${FS} \
+    --use_unet 1 \
+    --unet_path 'stae_2/output/unet.pt' \
+    --img_proj_path 'stage_2/output/img_proj.pt' \
+    --timestep_spacing 'uniform_trailing' \
+    --guidance_rescale 0.7 \
+    --perframe_ae
+