diff --git a/deploy/base_model/job_name_random_suffix.txt b/deploy/base_model/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..93a1847220f82253a91e2c686302c8416eee4a4a --- /dev/null +++ b/deploy/base_model/job_name_random_suffix.txt @@ -0,0 +1 @@ +r94995e diff --git a/deploy/base_model/submit_vbench2_generate_shards.sh b/deploy/base_model/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/base_model/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/base_model/vbench2_generate_shard_0.yaml b/deploy/base_model/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..38357ddc4e958d2ce70cf56b11f73c065360b79f --- /dev/null +++ b/deploy/base_model/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 USE_CHECKPOINT=false DOWNLOAD_CHECKPOINT=false SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/base_model/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh base_model 100" + + jobs: + - name: ywang29-vbench2-gen-base-r94995e-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/base_model/vbench2_generate_shard_1.yaml b/deploy/base_model/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ad8ec142707c720139612db680dcf3e186d7bd6c --- /dev/null +++ b/deploy/base_model/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 USE_CHECKPOINT=false DOWNLOAD_CHECKPOINT=false SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/base_model/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh base_model 100" + + jobs: + - name: ywang29-vbench2-gen-base-r94995e-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/base_model/vbench2_generate_shard_2.yaml b/deploy/base_model/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..43829b7db3032121ed5807177cd68c9071819292 --- /dev/null +++ b/deploy/base_model/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 USE_CHECKPOINT=false DOWNLOAD_CHECKPOINT=false SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/base_model/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh base_model 100" + + jobs: + - name: ywang29-vbench2-gen-base-r94995e-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/base_model/vbench2_generate_shard_3.yaml b/deploy/base_model/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fac692e5e38bdb770ea3e55a534b79ccf4f93186 --- /dev/null +++ b/deploy/base_model/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 USE_CHECKPOINT=false DOWNLOAD_CHECKPOINT=false SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/base_model/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh base_model 100" + + jobs: + - name: ywang29-vbench2-gen-base-r94995e-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/base_model/vbench2_generate_shard_4.yaml b/deploy/base_model/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0a34d4953dff586fc2f6b9f8cb3728197302423c --- /dev/null +++ b/deploy/base_model/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 USE_CHECKPOINT=false DOWNLOAD_CHECKPOINT=false SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/base_model/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh base_model 100" + + jobs: + - name: ywang29-vbench2-gen-base-r94995e-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/base_model/vbench2_generate_shard_5.yaml b/deploy/base_model/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9a2a46bff4fd1eb33745166d3c95c5d952a622fe --- /dev/null +++ b/deploy/base_model/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 USE_CHECKPOINT=false DOWNLOAD_CHECKPOINT=false SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/base_model/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh base_model 100" + + jobs: + - name: ywang29-vbench2-gen-base-r94995e-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/base_model/vbench2_generate_shard_6.yaml b/deploy/base_model/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..07d647c408b18d6422cb5960eecd1e72d4f70d68 --- /dev/null +++ b/deploy/base_model/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 USE_CHECKPOINT=false DOWNLOAD_CHECKPOINT=false SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/base_model/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh base_model 100" + + jobs: + - name: ywang29-vbench2-gen-base-r94995e-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/base_model/vbench2_generate_shard_7.yaml b/deploy/base_model/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0afef92090551a8f38f14cea8b910198f82d13cf --- /dev/null +++ b/deploy/base_model/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 USE_CHECKPOINT=false DOWNLOAD_CHECKPOINT=false SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/base_model/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh base_model 100" + + jobs: + - name: ywang29-vbench2-gen-base-r94995e-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/demo_vbench2_eval.yaml b/deploy/demo_vbench2_eval.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5674ff41f30115d472abf6faf2d1f589ab740082 --- /dev/null +++ b/deploy/demo_vbench2_eval.yaml @@ -0,0 +1,80 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" # "80" +# resource should scale accordingly to the gpu_num requested, cpu_num is recommended to be gpu_num x 11, same applies to the memory +cpu_num: "88" +memory: "880.0" + +# Add required resource and ENV vars for EFA (note: Only works on AWS, and only used for distributed training) +efa: False +# efa_count: "4" +# huge_pages: "21122Mi" + +priority: gpu-high + +pytorchjob: + replicas: 1 # number of replicas to run + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: omni-id + +# custom_node_labels: +# # use_case: research +# snap.com/zone: us-west-2c + + +script: +# pre_run_event: "cd /nfs/yli16/effvideo && pip install -r requirements.txt && pip uninstall -y apex && wandb login --relogin --host=https://snap.wandb.io local-5c7725e9c8a991d399dfee329e7c5c9f7b33de8e" + #pre_run_event: "cd /nfs/ywang29/video_reward/vila_ft && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19 && bash ./environment_setup.sh lift && source activate && conda activate lift && pip uninstall lift && pip install huggingface_hub==0.31.1 && pip install accelerate==0.32.0 && pip install peft==0.10.0" + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + + #pre_run_event: "bash ./environment_setup.sh lift && conda activate lift && pip uninstall lift && pip install huggingface_hub==0.31.1 " +# pre_run_event: "cd /nfs/yli16/snapgenv-v2-i2v && pip install -r requirements.txt && pip uninstall -y apex diffusers coremltools && pip install coremltools && wandb login --relogin --host=https://snap.wandb.io local-5c7725e9c8a991d399dfee329e7c5c9f7b33de8e" +# pre_run_event: "cd /nfs/yli16/snapgenv-v3 && pip install --upgrade pip && pip uninstall -y diffusers coremltools peft && pip install coremltools peft && wandb login --relogin --host=https://snap.wandb.io local-5c7725e9c8a991d399dfee329e7c5c9f7b33de8e" + # command: "cd /nfs/ywang29/Reward_finetuning/VideoX-Fun && source activate && source /nfs/ywang29/Reward_finetuning/VideoX-Fun/scripts/wan2.1/env.sh && /nfs/ywang29/Reward_finetuning/VideoX-Fun/validation_full_vbench2.0_control.sh && while :; do sleep 1000; done" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto bash run_sage_vbench2_generate.sh yw004-flow-baseline 100 && RUN_SETUP=true bash run_vbench2_eval_only.sh yw004-flow-baseline 100" + + + jobs: + - name: ywang29-vbench2-eval-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +#docker_image: 440036398022.dkr.ecr.us-west-2.amazonaws.com/facecraft-ml:efficient_video_ywu6-6b32aee +#docker_image: 440036398022.dkr.ecr.us-west-2.amazonaws.com/facecraft-ml:efficient_video_ywu6-3942a73 +# docker_image: 440036398022.dkr.ecr.us-west-2.amazonaws.com/facecraft-ml:efficient_video_ywu6-eb1a4a7 +# docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.7.1-gpu-py312-cu128-ubuntu22.04-ec2 +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + #prefix: logs/ # optional bucket's prefix + #region: us-west-2 # optional bucket's region (by default cluster's region will be used) + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 # generative-one-shot + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 + + +#mkdir: /nfs/ywu6/logs \ No newline at end of file diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/job_name_random_suffix.txt b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..98fc6bfdc0bce4716b822e6527e6801aaab98301 --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/job_name_random_suffix.txt @@ -0,0 +1 @@ +r9d1839 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/submit_vbench2_generate_shards.sh b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_0.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7285f0e71ab75b9585913b4ce5a3c2d0269f42ee --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flowsw4-r9d1839-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_1.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..47051651af87fd98c32c605003c8c90c81af5d51 --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flowsw4-r9d1839-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_2.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4261e5f8e551b539381970731c7a3ddc0dde57a1 --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flowsw4-r9d1839-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_3.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..6cd4b01ec7874ee9f8ec454a58b0dca28e797310 --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flowsw4-r9d1839-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_4.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7713e15fc6d67f6c112ac5f1bc9d4401c27cbb86 --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flowsw4-r9d1839-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_5.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8a80321c2f576ef1cbe50077b824400fe15fe5fe --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flowsw4-r9d1839-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_6.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..13d1e0e6c01cf84a10103e760cd0e7d897ed9861 --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flowsw4-r9d1839-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_7.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8d743825834936d7afd62ab3c3e7764f12886466 --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flowsw4-r9d1839-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/job_name_random_suffix.txt b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..a0dec31dc54533cea0d8e3acc0852fc903becf10 --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/job_name_random_suffix.txt @@ -0,0 +1 @@ +raea4fc diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/submit_vbench2_generate_shards.sh b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_0.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f1db2a8c5c21fd7c676eac152a652db39627197e --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv1 100" + + jobs: + - name: ywang29-vbench2-gen-flowadv1-raea4fc-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_1.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4831bd7ab4e73563f1dabae9bb4cd8e69774f00d --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv1 100" + + jobs: + - name: ywang29-vbench2-gen-flowadv1-raea4fc-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_2.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..de7384937410f9b95f531575bac589a75d152f50 --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv1 100" + + jobs: + - name: ywang29-vbench2-gen-flowadv1-raea4fc-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_3.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..159ff35a4bb7e1a4c8ec65c3d80fb04cc5691285 --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv1 100" + + jobs: + - name: ywang29-vbench2-gen-flowadv1-raea4fc-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_4.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8fdf436d64b98566de4bcce9be04fbc3f7b28fdc --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv1 100" + + jobs: + - name: ywang29-vbench2-gen-flowadv1-raea4fc-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_5.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..21e20f03e18c8ed33ff4d7e00152962e1089dd77 --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv1 100" + + jobs: + - name: ywang29-vbench2-gen-flowadv1-raea4fc-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_6.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9644e9fd7da429a842f69af0e6401c70a2c2c6af --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv1 100" + + jobs: + - name: ywang29-vbench2-gen-flowadv1-raea4fc-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_7.yaml b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..259e91c3ee7c5ceec2169167b78562d3e378143b --- /dev/null +++ b/deploy/jun16/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv1 100" + + jobs: + - name: ywang29-vbench2-gen-flowadv1-raea4fc-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/job_name_random_suffix.txt b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..8462ac2045282fc3834413028e679065b8cb56c5 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/job_name_random_suffix.txt @@ -0,0 +1 @@ +r4f22fc diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/submit_vbench2_generate_shards.sh b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_0.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d0eadc9423c432c5963931b4b8547ec4944ba16a --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcva4-r4f22fc-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_1.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a77c2006cc6ec916f38751e5042435f4105bc290 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcva4-r4f22fc-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_2.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ea5380f2eca3234045fa24fa6866cc1e3677f85a --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcva4-r4f22fc-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_3.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0a156e682251b7945c79c10a9a85ad794cacd2ac --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcva4-r4f22fc-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_4.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..875f7b66f47d43bd37d4245b0069ab462632c2e5 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcva4-r4f22fc-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_5.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1c7dc73278fb0055af7b86ce35c30eda2f96df05 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcva4-r4f22fc-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_6.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b8feaff8f713ab0a7d3f1dfb865bd590a0eb3041 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcva4-r4f22fc-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_7.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7863f0577aea6485f3409ef112b4a47bbd476f3b --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcva4-r4f22fc-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/job_name_random_suffix.txt b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..b0305af02d667c8d774ff5ac5f2494d92c10d20b --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/job_name_random_suffix.txt @@ -0,0 +1 @@ +rc305a9 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/submit_vbench2_generate_shards.sh b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_0.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a4ad121163a0630581be60c8e837a329eb1bfe55 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-chk4-rc305a9-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_1.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a4b448eceb40d5a6a44148e330af95572c1c0624 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-chk4-rc305a9-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_2.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..68bc62ab97c6b9acba2bd89fbe4e091144670f8c --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-chk4-rc305a9-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_3.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9da6bbb6122838988a37466373e314b939d91c7f --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-chk4-rc305a9-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_4.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..34def5601c009c9779216f42189911fd5594be03 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-chk4-rc305a9-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_5.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..dac28c499919729f2f2f5c92e7158859dddcb733 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-chk4-rc305a9-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_6.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e3bf832effce75a7dcbd2e9f75af6d951724e6db --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-chk4-rc305a9-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_7.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..531c9f33a1243321b5e50202a2809288429cb7c8 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-chk4-rc305a9-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/job_name_random_suffix.txt b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..e2373b45811ba3cbf92c91280cdeb2f55da75fb9 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/job_name_random_suffix.txt @@ -0,0 +1 @@ +r93d318 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/submit_vbench2_generate_shards.sh b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_0.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..250406d69d859ed3d1c4299ea80ae7cdafa151d0 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-finemap-r93d318-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_1.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0b707f45b2755ac501269f012d4246540750f4dd --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-finemap-r93d318-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_2.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7f16b1bf083016e107009f7240fb87eb16e9d785 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-finemap-r93d318-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_3.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..32ec20beae769231243c84438491de256c7de565 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-finemap-r93d318-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_4.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..354007c03cf22b7448aefc7175e8c17ca3dd7675 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-finemap-r93d318-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_5.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..51839e50e2f7fae42fe7fd18e3d2feb2ae105d8c --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-finemap-r93d318-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_6.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..cd8eb193c8eea5d21ec280b205b4783cc4c714dc --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-finemap-r93d318-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_7.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..50cb42ed18483a400e5d0291a1c12b47da5b8069 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-finemap-r93d318-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/job_name_random_suffix.txt b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..4a97203ac3b80737ea5e0f4fa1deb717cfcd178c --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/job_name_random_suffix.txt @@ -0,0 +1 @@ +rcf2e90 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/submit_vbench2_generate_shards.sh b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_0.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4d791715beaadce26f2e3c12ad544a2ba0ff75d2 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-win3-rcf2e90-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_1.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4e7e9d8ad46e83aa566df998a2fdf57378f7245a --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-win3-rcf2e90-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_2.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1640301ba23d6500d26db7298c5a53718ab895c2 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-win3-rcf2e90-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_3.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..125182d8689d0bf506d3de22c124bc609078d07d --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-win3-rcf2e90-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_4.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e1c3fba636a63f9f208773fa8c12a25d16974074 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-win3-rcf2e90-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_5.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bb0c5d11127edf22ea9c55f8ca4f28648f3a7fa5 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-win3-rcf2e90-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_6.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b0429809ef297fd9f60309887b9df5e07dc5f97c --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-win3-rcf2e90-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_7.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a028a0fe57f8f58c14fcc3acaa4ac0d3d3ff83f4 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-win3-rcf2e90-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/job_name_random_suffix.txt b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..f34a8223b600aa4ff3b075db33e01be1bb604fda --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/job_name_random_suffix.txt @@ -0,0 +1 @@ +r019ef9 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/submit_vbench2_generate_shards.sh b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_0.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1ba5d6c3aa2589d229781b821db7f82e4b31de39 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-win5-r019ef9-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_1.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9bdfc0bc421830a0c2a2b403a2a5b0684d9576d2 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-win5-r019ef9-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_2.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f1e1e7e7eacb2a883ad4985caf6abf487c4d437e --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-win5-r019ef9-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_3.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8842abf1979ba9de400cb8a297da17ec6a2e450d --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-win5-r019ef9-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_4.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bf875c4ef05738c0c3dd24c069ceacc4310a5872 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-win5-r019ef9-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_5.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ff9bc76a1128d469204ef37cc9bc5ad4330af21b --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-win5-r019ef9-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_6.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..da63834a013be16a65e1df67e842c5eccb39091b --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-win5-r019ef9-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_7.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..beb73caf941cc996d9c22ee93c19fd40f8c5b784 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-win5-r019ef9-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/job_name_random_suffix.txt b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..a041b6c411b6528dad3a27579b45b785736de86c --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/job_name_random_suffix.txt @@ -0,0 +1 @@ +r190b75 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/submit_vbench2_generate_shards.sh b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_0.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..50b342b81f2816887fb4c0c2a8c59c2710dcceba --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-win7-r190b75-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_1.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f6e1dfc3f4b652392b46f90ef274c9ec219a7956 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-win7-r190b75-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_2.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..07427f0272a6cac7a2341e4b3f022c553c4e9be8 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-win7-r190b75-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_3.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9eeab3537115fb78d379b8ec0d29680935beccef --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-win7-r190b75-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_4.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c5a0489a21ff40842471f20dc1199d85a7737c84 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-win7-r190b75-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_5.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c7356d880918e56b13ddf8b1be0c2b7ba86aaa5c --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-win7-r190b75-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_6.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..91a77019bbee8412b7f61a491c6d4390e88e2c8f --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-win7-r190b75-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_7.yaml b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fa3ef3272a01ee86315e1568c34f5d6437cad539 --- /dev/null +++ b/deploy/jun20/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-win7-r190b75-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/job_name_random_suffix.txt b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..c9a26252f6be6e7a964ae84f3f30bd2e9e4fb891 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/job_name_random_suffix.txt @@ -0,0 +1 @@ +rf26da3 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/submit_vbench2_generate_shards.sh b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_0.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..50f2776b999139391894ebc69005a9a3b7442677 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-vs21c4-rf26da3-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_1.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fa0fca5237fb9f46d3943707b2b8f8456f66015f --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-vs21c4-rf26da3-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_2.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5b112928c95286f9227591c0cbf0c353fa78f1b2 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-vs21c4-rf26da3-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_3.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..421f0a93144e92fed660f0335d443dc6081f777f --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-vs21c4-rf26da3-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_4.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..994a40dd2d763eea4b17e4ebfeb0a89365930423 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-vs21c4-rf26da3-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_5.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..77825bd29c74f6c5051e3480568c9f9589d17364 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-vs21c4-rf26da3-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_6.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..49e48a91d3bee489048eaf5769a7ac17d17bef43 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-vs21c4-rf26da3-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_7.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9c01f074b18069e814c3c92dd2b50ddb8fac969d --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-chunk4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-chunk4 100" + + jobs: + - name: ywang29-vbench2-gen-vs21c4-rf26da3-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/job_name_random_suffix.txt b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..21d4357ca24d3fd5266a413beca3c009eecaf2cd --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/job_name_random_suffix.txt @@ -0,0 +1 @@ +r485620 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/submit_vbench2_generate_shards.sh b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_0.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..36308eb26641bc5d9d71fec5e1619c6651e29e7d --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-vs21fm-r485620-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_1.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1a621f656c23c279c6ad6fcfbae79a5e1b694d50 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-vs21fm-r485620-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_2.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..756d699affa0d4d98ecda4a7ec8046e1f09be572 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-vs21fm-r485620-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_3.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a536cfae30316f436655335975d4a332757395b9 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-vs21fm-r485620-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_4.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..582103208d2c6617602a6a2de5d9ad4344c2cc67 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-vs21fm-r485620-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_5.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..37ac23bdbb9e067d927f3798073b68d937601257 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-vs21fm-r485620-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_6.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4400d7a66e399e96bc349b35886f4c3bf7f1a891 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-vs21fm-r485620-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_7.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8b9172f82f1544172ec6f55ae5fcbc646cd93a5d --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-fine-map/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-fine-map 100" + + jobs: + - name: ywang29-vbench2-gen-vs21fm-r485620-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/job_name_random_suffix.txt b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..9f8e849a90c1420d4b1b9adebeb68e9d812b1e77 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/job_name_random_suffix.txt @@ -0,0 +1 @@ +rc571fc diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/submit_vbench2_generate_shards.sh b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_0.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..6b5f2b2cb0314ccd0087285231bb7953535a1640 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w3-rc571fc-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_1.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..468d4119f3ae18c3b0188c9ca698f7178ccfcf31 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w3-rc571fc-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_2.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..84b39a3f3f49e67e86055479745ba25b13cfc36a --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w3-rc571fc-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_3.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..3dff33c99766b971efb05ef838e376106cce7e15 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w3-rc571fc-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_4.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9ce481810004fa05b4d8b4d2bb19eac6f9a44627 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w3-rc571fc-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_5.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5a6af11dced9eef5218e995edd2585d446de409e --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w3-rc571fc-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_6.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..40539e974c9d15e267dfe7f85789bc55f73e3c89 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w3-rc571fc-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_7.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..19e0b70521b71aca471f0fca5daca64a9d8b863b --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window3 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w3-rc571fc-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/job_name_random_suffix.txt b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..6a6256b10745942aea451e6ee3d4ab13b6ad2f40 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/job_name_random_suffix.txt @@ -0,0 +1 @@ +rec5ea4 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/submit_vbench2_generate_shards.sh b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_0.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7335d703355e0c4bc050f1e2e847a0dc99d5570b --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w5-rec5ea4-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_1.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..2a99176f4b85834d8ec2d332cb16e378698a4ea1 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w5-rec5ea4-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_2.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..dd3c1a4a4a122972da2d1411f8b733bb7311d964 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w5-rec5ea4-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_3.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a771aec38541fc13732d030066b5cec6a836ee0f --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w5-rec5ea4-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_4.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..43edb97e44e855c45359a009dbae9b6449ee319f --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w5-rec5ea4-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_5.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0afcfc8586a5c63acce0f8444ae10f353e037e2d --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w5-rec5ea4-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_6.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..025931f39ee85bd51b0d1cac380defa91755d9a7 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w5-rec5ea4-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_7.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..dd341b13bf43b7c33056a5539050c5bdb86bd8a3 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window5 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w5-rec5ea4-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/job_name_random_suffix.txt b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..74f6b8b212ff3bcd9e602a4d7fedf1a6f7c162a8 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/job_name_random_suffix.txt @@ -0,0 +1 @@ +r64494e diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/submit_vbench2_generate_shards.sh b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_0.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f19f9c539e5818fe5850a622954e6250932d60fe --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w7-r64494e-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_1.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..84cc90cc0a8382944a2d5ee624e1f4c878a675ac --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w7-r64494e-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_2.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7d6238a99ab62c33f509dbdb6b822126bd99ecfe --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w7-r64494e-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_3.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0a7f9bd1fdb4dffbe3a217d7c18a30cdb3826458 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w7-r64494e-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_4.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..070a49143cd06c0b84da7c02163ceb6b86b08c0a --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w7-r64494e-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_5.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..658d4d30439cf18ff6b1725297ccc5810cc1ec64 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w7-r64494e-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_6.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f5f8b7275ef17f2dcf35426c7062ce6376c25f38 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w7-r64494e-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_7.yaml b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7239a82f1a101d689f81e4772ee3cd582d5161a0 --- /dev/null +++ b/deploy/jun20/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv-window7/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv-window7 100" + + jobs: + - name: ywang29-vbench2-gen-vs21w7-r64494e-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/job_name_random_suffix.txt b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..a4b9c4b2e69d9ff129d7876ed93318067d7e6fd8 --- /dev/null +++ b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/job_name_random_suffix.txt @@ -0,0 +1 @@ +r8cba40 diff --git a/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/submit_vbench2_generate_shards.sh b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_0.yaml b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..12bb1dedb053935ebb229e78cf32fb8bfee35e94 --- /dev/null +++ b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-cpsnw-rc3e22a-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_1.yaml b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e76aec9907f247e130fb07154bc547f8edcc5d87 --- /dev/null +++ b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-cpsnw-rc3e22a-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_2.yaml b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ee47deb2ff191e61bba03ed3b560308e2a60e7d7 --- /dev/null +++ b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-cpsnw-rc3e22a-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_3.yaml b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e2cd0c80ab6a96b7e12a5b5e8bf9c3dcc679de0c --- /dev/null +++ b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-cpsnw-rc3e22a-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_4.yaml b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..cb8cd7d8087fd514deb0c845fd0ff499ee39935a --- /dev/null +++ b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-cpsnw-rc3e22a-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_5.yaml b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fcc72cf7af074c0ba6044ff93926d7ca41fe4049 --- /dev/null +++ b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-cpsnw-rc3e22a-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_6.yaml b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..521e01a6221d6e084fd9b932143637ca782507b2 --- /dev/null +++ b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-cpsnw-rc3e22a-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_7.yaml b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b96f3ec5df672ab1a844a4c620778e3620f14835 --- /dev/null +++ b/deploy/jun9/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-cpsnw-rc3e22a-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/job_name_random_suffix.txt b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..a4b9c4b2e69d9ff129d7876ed93318067d7e6fd8 --- /dev/null +++ b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/job_name_random_suffix.txt @@ -0,0 +1 @@ +r8cba40 diff --git a/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/submit_vbench2_generate_shards.sh b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_0.yaml b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a21347e84252aa688a1eb1b6d65d963584864d4e --- /dev/null +++ b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flwnw4-rf2bcff-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_1.yaml b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5c61fd109294bebd78f95ab1cfc425ae02c5d399 --- /dev/null +++ b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flwnw4-rf2bcff-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_2.yaml b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7ddf037c0d6e2b8d056f7b8313cd44eabb1b61e1 --- /dev/null +++ b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flwnw4-rf2bcff-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_3.yaml b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4c15baae539d26d3ebd61003109720ac12fef536 --- /dev/null +++ b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flwnw4-rf2bcff-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_4.yaml b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..19b9500eeee84370efd465f727e96db01dd1224a --- /dev/null +++ b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flwnw4-rf2bcff-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_5.yaml b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..472b79ea049d66bcfde56a0fefc8957f61a938ef --- /dev/null +++ b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flwnw4-rf2bcff-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_6.yaml b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c95a749120d1eb44bb442b04dd7a6a8b112c9757 --- /dev/null +++ b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flwnw4-rf2bcff-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_7.yaml b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..26c37b4514844bc25820a3c783d0a7471f7435d9 --- /dev/null +++ b/deploy/jun9/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-flwnw4-rf2bcff-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/job_name_random_suffix.txt b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..3e3de89b4fc231f6b300f071b9da1ad0b87d4971 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/job_name_random_suffix.txt @@ -0,0 +1 @@ +g4e4b1-r8a6dd9 diff --git a/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/submit_vbench2_generate_shards.sh b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_0.yaml b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ffa3504e95a56365ba5415bff1c5060bf628234d --- /dev/null +++ b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b1-r8a6dd9-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_1.yaml b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..3446c4ac780691ff0d2f467a7d6355188eda7784 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b1-r8a6dd9-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_2.yaml b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4cceaf951a953f0a044e7b57d03bdf087508df30 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b1-r8a6dd9-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_3.yaml b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..6955b46e389011c688ec3e440152b863828ec4e1 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b1-r8a6dd9-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_4.yaml b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..3557ecec818acae36d1f37ea7dbaab48c3dc0e03 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b1-r8a6dd9-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_5.yaml b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..679d8774b36abb8d7df277a5c4b763d4953a7e09 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b1-r8a6dd9-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_6.yaml b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..994452385a1b67cc00b7df68c245f4f71b58e557 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b1-r8a6dd9-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_7.yaml b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..538732f4a61f2302b0efbeb8e77bbb6f653cbd4d --- /dev/null +++ b/deploy/jun9/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b1-r8a6dd9-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/job_name_random_suffix.txt b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..dc0e906fc433902542bca1a3274ab99a2e27af98 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/job_name_random_suffix.txt @@ -0,0 +1 @@ +fr30-r20b41a diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/submit_vbench2_generate_shards.sh b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_0.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d62cd6cb0b2c1bb0ae246fe9c933047622ae3c24 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30 100" + + jobs: + - name: ywang29-vbench2-gen-fr30-r20b41a-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_1.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bbbb23c972cd6a3a6ab791ac7f0b7092fd2a793f --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30 100" + + jobs: + - name: ywang29-vbench2-gen-fr30-r20b41a-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_2.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..48e7af1a2f5c4fecbacf8e4ef047c36891ce30c4 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30 100" + + jobs: + - name: ywang29-vbench2-gen-fr30-r20b41a-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_3.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..29166715f555cc604bf54eb2764e6cdf721382f1 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30 100" + + jobs: + - name: ywang29-vbench2-gen-fr30-r20b41a-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_4.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b079b314dc495fe80b1e827aab1870269180f2c2 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30 100" + + jobs: + - name: ywang29-vbench2-gen-fr30-r20b41a-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_5.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..56dd9f2a3b30168da5d51b2fad4c63e1af782a19 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30 100" + + jobs: + - name: ywang29-vbench2-gen-fr30-r20b41a-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_6.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..cfd116a5f9aa959ef24e9612760defd16754fa1b --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30 100" + + jobs: + - name: ywang29-vbench2-gen-fr30-r20b41a-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_7.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f831cf24c202a8d2a3f66cf62b9fa9092827206e --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30 100" + + jobs: + - name: ywang29-vbench2-gen-fr30-r20b41a-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/job_name_random_suffix.txt b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..d767c4851441ff0d6cf235cfa2e9a5650a0d83ca --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/job_name_random_suffix.txt @@ -0,0 +1 @@ +rc403c0 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/submit_vbench2_generate_shards.sh b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_0.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..85fc8a482339675de7015011a770b4ce49cb0432 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5 100" + + jobs: + - name: ywang29-vbench2-gen-fr5-rc403c0-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_1.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9c8300e269a04d59e8e77cddcd6c6220a85ba55d --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5 100" + + jobs: + - name: ywang29-vbench2-gen-fr5-rc403c0-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_2.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d16934f743db427e185bbe871cce9b19d59a4128 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5 100" + + jobs: + - name: ywang29-vbench2-gen-fr5-rc403c0-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_3.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..6ea49c255488c29b99fba110ca808756074cdbd9 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5 100" + + jobs: + - name: ywang29-vbench2-gen-fr5-rc403c0-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_4.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4d6926d517edd7b84309e16780a572e03573c7d2 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5 100" + + jobs: + - name: ywang29-vbench2-gen-fr5-rc403c0-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_5.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..126d9a040e7de4209c158d061063b0331416414b --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5 100" + + jobs: + - name: ywang29-vbench2-gen-fr5-rc403c0-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_6.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..710f9b08754725cc49d62efecc8fee382f2fb59f --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5 100" + + jobs: + - name: ywang29-vbench2-gen-fr5-rc403c0-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_7.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..2c888f4979b5e352cbd1099b0fe41782e24aaa9e --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl5 100" + + jobs: + - name: ywang29-vbench2-gen-fr5-rc403c0-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/job_name_random_suffix.txt b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..23d46fb2465a1757cfea21dfeb692b5c49ceec57 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/job_name_random_suffix.txt @@ -0,0 +1 @@ +r72cdfa diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/submit_vbench2_generate_shards.sh b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_0.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..37b33d4e0260c1901f9523b4ea9a69d407be0b68 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5 100" + + jobs: + - name: ywang29-vbench2-gen-rt5-r72cdfa-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_1.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..6e21218d794d0a15e21a81fb3910c0571c5b4c8c --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5 100" + + jobs: + - name: ywang29-vbench2-gen-rt5-r72cdfa-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_2.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0fb314fa1700debdfc90450c5b3e544d5b9decab --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5 100" + + jobs: + - name: ywang29-vbench2-gen-rt5-r72cdfa-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_3.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b5ffbdbfdf5a3433da0de79e79d8b86e9f25e55c --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5 100" + + jobs: + - name: ywang29-vbench2-gen-rt5-r72cdfa-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_4.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b071e33c1c5c274e982e848faa287a0b8c51cf26 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5 100" + + jobs: + - name: ywang29-vbench2-gen-rt5-r72cdfa-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_5.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e8f7e108537e52d6c60f6faf27e8819df6a94500 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5 100" + + jobs: + - name: ywang29-vbench2-gen-rt5-r72cdfa-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_6.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..2e6c9fc6c14af075c5726299559bf38655fc3995 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5 100" + + jobs: + - name: ywang29-vbench2-gen-rt5-r72cdfa-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_7.yaml b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ee311feff70a7d397bb4d2bad2b43c6a1a2168b0 --- /dev/null +++ b/deploy/jun9/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5 100" + + jobs: + - name: ywang29-vbench2-gen-rt5-r72cdfa-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/job_name_random_suffix.txt b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..ede6b8e7ac8aab02d02f0d0a11457472260eeed6 --- /dev/null +++ b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/job_name_random_suffix.txt @@ -0,0 +1 @@ +ynm-r2f8168 diff --git a/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/submit_vbench2_generate_shards.sh b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_0.yaml b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..25fec17538b226cc54b0c8fd7a8d4796cc863473 --- /dev/null +++ b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-sage-qwen35-9b-yes-no-margin/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-sage-qwen35-9b-yes-no-margin 100" + + jobs: + - name: ywang29-vbench2-gen-ynm-r2f8168-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_1.yaml b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8c5171092b0b2e7e6f1812640415ad7def1d0aea --- /dev/null +++ b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-sage-qwen35-9b-yes-no-margin/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-sage-qwen35-9b-yes-no-margin 100" + + jobs: + - name: ywang29-vbench2-gen-ynm-r2f8168-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_2.yaml b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..47e548b2beaef4525862070bdaf55218867f7d2b --- /dev/null +++ b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-sage-qwen35-9b-yes-no-margin/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-sage-qwen35-9b-yes-no-margin 100" + + jobs: + - name: ywang29-vbench2-gen-ynm-r2f8168-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_3.yaml b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7c9810b1b76c077be84a780aaf9d7aa6b72df0bb --- /dev/null +++ b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-sage-qwen35-9b-yes-no-margin/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-sage-qwen35-9b-yes-no-margin 100" + + jobs: + - name: ywang29-vbench2-gen-ynm-r2f8168-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_4.yaml b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1fa7e72ca848713093403310b996814997169c6b --- /dev/null +++ b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-sage-qwen35-9b-yes-no-margin/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-sage-qwen35-9b-yes-no-margin 100" + + jobs: + - name: ywang29-vbench2-gen-ynm-r2f8168-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_5.yaml b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..54669072d80cf5fbf7a4c94e2b82b03f16352d26 --- /dev/null +++ b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-sage-qwen35-9b-yes-no-margin/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-sage-qwen35-9b-yes-no-margin 100" + + jobs: + - name: ywang29-vbench2-gen-ynm-r2f8168-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_6.yaml b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..51fac2f8b949a022361c92677dcad28e4eac18ef --- /dev/null +++ b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-sage-qwen35-9b-yes-no-margin/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-sage-qwen35-9b-yes-no-margin 100" + + jobs: + - name: ywang29-vbench2-gen-ynm-r2f8168-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_7.yaml b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..edb13ce3cacb6857b2c5e3fdaf08179fbb22eea7 --- /dev/null +++ b/deploy/jun9/yw006-sage-qwen35-9b-yes-no-margin/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-sage-qwen35-9b-yes-no-margin/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-sage-qwen35-9b-yes-no-margin 100" + + jobs: + - name: ywang29-vbench2-gen-ynm-r2f8168-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/job_name_random_suffix.txt b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..6de86f3b8b3023896d0a3fcd82279cdcd22fe6b1 --- /dev/null +++ b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/job_name_random_suffix.txt @@ -0,0 +1 @@ +r73eb73 diff --git a/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/submit_vbench2_generate_shards.sh b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_0.yaml b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..698df7fddb5383eb95496569e4ef98e55a089f42 --- /dev/null +++ b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_1.yaml b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5e01485c55ba920529e34fd638e9a54ad579cd05 --- /dev/null +++ b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_2.yaml b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..78939870a6615646f279eccf3843ec56e0d30bdb --- /dev/null +++ b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_3.yaml b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9a21e4eeaa2463fdf529f93d2c05bd518b1eac69 --- /dev/null +++ b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_4.yaml b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b85fe1f89d63be7c151c37e26f9e14fdcef1bc03 --- /dev/null +++ b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_5.yaml b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c07cfab6077436278f2c10469ea343c7139ac521 --- /dev/null +++ b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_6.yaml b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..2a5e4a72d918a03c4f007ccd5db189def0d1e84a --- /dev/null +++ b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_7.yaml b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c345a5c5d5e7db845d02618cf7de9372d71c3538 --- /dev/null +++ b/deploy/jun9/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: diffusion-training-acceleration +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/job_name_random_suffix.txt b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..a4b9c4b2e69d9ff129d7876ed93318067d7e6fd8 --- /dev/null +++ b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/job_name_random_suffix.txt @@ -0,0 +1 @@ +r8cba40 diff --git a/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/submit_vbench2_generate_shards.sh b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_0.yaml b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4eff0ed4bd37e40ccc6a64b79a787e2e6791482a --- /dev/null +++ b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-nw4-r8cba40-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_1.yaml b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..749c4b0c39e6781458cde4e47a942010af5e2f6b --- /dev/null +++ b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-nw4-r8cba40-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_2.yaml b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9706f3213f0fc7d8e9f181504eac164970231fb0 --- /dev/null +++ b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-nw4-r8cba40-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_3.yaml b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..eb64083895cfc914b53c801c555dbe43565cbdba --- /dev/null +++ b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-nw4-r8cba40-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_4.yaml b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..766499deb8c9e4c3532f8da7807b7838ee412719 --- /dev/null +++ b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-nw4-r8cba40-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_5.yaml b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..894b1a0dddb2e2901f6d2b3c7c114f4b99c02765 --- /dev/null +++ b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-nw4-r8cba40-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_6.yaml b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..99213052ca66ed2401d51b896bb107ee80e8ce95 --- /dev/null +++ b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-nw4-r8cba40-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_7.yaml b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1e35f97c63064d2ac54e99fc5839b7698aef4f66 --- /dev/null +++ b/deploy/jun_5/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-nw4-r8cba40-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/job_name_random_suffix.txt b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..b4f1b5b6c99285fc56abac58084b6a252516900a --- /dev/null +++ b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/job_name_random_suffix.txt @@ -0,0 +1 @@ +re235af diff --git a/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/submit_vbench2_generate_shards.sh b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_0.yaml b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c0c1b064f5ab054c60883393d7d120894dd67cf4 --- /dev/null +++ b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw007-1-sage-videoalign-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-va1-re235af-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_1.yaml b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e4326a463d75ff49480255a6f9f113e5a8e9959d --- /dev/null +++ b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw007-1-sage-videoalign-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-va1-re235af-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_2.yaml b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..854382c35641fa31e9cf82da4b6dacdeb19af80c --- /dev/null +++ b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw007-1-sage-videoalign-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-va1-re235af-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_3.yaml b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bbe85c3108d27de3dee9fffc4b62f12464a7ea13 --- /dev/null +++ b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw007-1-sage-videoalign-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-va1-re235af-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_4.yaml b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..12086cd57898f18ce4d690f5bdcfb588f9aed9e7 --- /dev/null +++ b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw007-1-sage-videoalign-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-va1-re235af-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_5.yaml b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..92a732200549ccaf2a8692f231b252d6c6160df0 --- /dev/null +++ b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw007-1-sage-videoalign-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-va1-re235af-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_6.yaml b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f62df5091536a4a418c79cfeef2aa6d859da4608 --- /dev/null +++ b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw007-1-sage-videoalign-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-va1-re235af-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_7.yaml b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4616d9daed41332a6a56b762f137a3a6132a8148 --- /dev/null +++ b/deploy/jun_5/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw007-1-sage-videoalign-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw007-1-sage-videoalign-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-va1-re235af-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/job_name_random_suffix.txt b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..7ef8a3693461b6226afb578c6ba6ccd40d6316b1 --- /dev/null +++ b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/job_name_random_suffix.txt @@ -0,0 +1 @@ +r60062c diff --git a/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/submit_vbench2_generate_shards.sh b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_0.yaml b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..779c127c0c8b8d6b94dbca57f977c84ed0ef2560 --- /dev/null +++ b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs21-r60062c-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_1.yaml b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e2108f4330b5bef10fed9599552185d3f0d7ff90 --- /dev/null +++ b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs21-r60062c-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_2.yaml b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a7ea267668046836e1e0a2aef88e440c4a85f4fb --- /dev/null +++ b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs21-r60062c-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_3.yaml b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..895a0266441d7e4d494a85cbb469f44b95099fbf --- /dev/null +++ b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs21-r60062c-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_4.yaml b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..32018cf65024d13bbfda5ee6017faa424b67e203 --- /dev/null +++ b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs21-r60062c-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_5.yaml b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a2374e702ee2708c7fa1266b3d05056d10427f2a --- /dev/null +++ b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs21-r60062c-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_6.yaml b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e6e15651d6fad2ea1be28ba001d1b4d9563926ab --- /dev/null +++ b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs21-r60062c-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_7.yaml b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..df660698f4071e8290ac8854264518a4dfdc6f76 --- /dev/null +++ b/deploy/jun_5/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs21-r60062c-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/job_name_random_suffix.txt b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..3bcd43fd32fe276fb9d7fcb06c24e1eef293fae6 --- /dev/null +++ b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/job_name_random_suffix.txt @@ -0,0 +1 @@ +r8fe0e6 diff --git a/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/submit_vbench2_generate_shards.sh b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_0.yaml b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7d9300bde83dd48f443633e624adaa7677e70a93 --- /dev/null +++ b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs22nw-r8fe0e6-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_1.yaml b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bf309af95ae4f6bc6682f90b2c321e3c19cc63d0 --- /dev/null +++ b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs22nw-r8fe0e6-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_2.yaml b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..3b2f2a8259e0dfafa3c774b165f8c2fd21acc1f0 --- /dev/null +++ b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs22nw-r8fe0e6-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_3.yaml b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b5341bdc213959a3628c1310092be3481f34fba8 --- /dev/null +++ b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs22nw-r8fe0e6-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_4.yaml b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..76118f8dff66155d6b58bda846c5ff6474cabb8e --- /dev/null +++ b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs22nw-r8fe0e6-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_5.yaml b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d665ff796039152c7d4774b50b697caa2acac276 --- /dev/null +++ b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs22nw-r8fe0e6-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_6.yaml b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8228c5e09f966d02ad318157cb6aa2b084b43830 --- /dev/null +++ b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs22nw-r8fe0e6-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_7.yaml b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..3da8b890fc42f95c0bc528ed0d02df1d948032f8 --- /dev/null +++ b/deploy/jun_5/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-2-sage-videoscore2-token-credit-video-adv-no-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs22nw-r8fe0e6-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/submit_vbench2_generate_shards.sh b/deploy/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..8352139b4fcffc2d420e6ad3e9d1676004ca8da0 --- /dev/null +++ b/deploy/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-16}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/vbench2_eval_yw002_1_sage_baseline.yaml b/deploy/vbench2_eval_yw002_1_sage_baseline.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8fe62048b4c439a5946a7b3845150e34d13f7a4a --- /dev/null +++ b/deploy/vbench2_eval_yw002_1_sage_baseline.yaml @@ -0,0 +1,53 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone + +script: + pre_run_event: "apt-get update -y && apt-get install -y inetutils-ping rsync ffmpeg gpustat && (cp /nfs/ywang29/.aws ~ -r || true) && ulimit -c 0" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO && RUN_SETUP=true RUN_SMOKE_TEST=true RUN_EVAL=true EVAL_MODE=standard EVAL_CUDA_VISIBLE_DEVICES=auto DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw002-1-sage-baseline/vbench2_eval/step_100 STAGED_VIDEO_ROOT=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw002-1-sage-baseline/vbench2_eval/step_100/videos EVAL_OUTPUT_ROOT=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw002-1-sage-baseline/vbench2_eval/step_100/vbench2_results SELECTED_INFO_JSON=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw002-1-sage-baseline/vbench2_eval/step_100/vbench2_selected_info.json bash run_vbench2_eval_only.sh yw002-1-sage-baseline 100" + + jobs: + - name: ywang29-vbench2-eval-yw002-1-sage-baseline + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_eval_yw006_4_token_credit_video_adv_2.yaml b/deploy/vbench2_eval_yw006_4_token_credit_video_adv_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c086532796f654b5f33e2ef637613cd70b0c6c07 --- /dev/null +++ b/deploy/vbench2_eval_yw006_4_token_credit_video_adv_2.yaml @@ -0,0 +1,53 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone + +script: + pre_run_event: "apt-get update -y && apt-get install -y inetutils-ping rsync ffmpeg gpustat && (cp /nfs/ywang29/.aws ~ -r || true) && ulimit -c 0" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO && RUN_SETUP=true RUN_SMOKE_TEST=true RUN_EVAL=true EVAL_MODE=standard EVAL_CUDA_VISIBLE_DEVICES=auto DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/vbench2_eval/step_100 STAGED_VIDEO_ROOT=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/vbench2_eval/step_100/videos EVAL_OUTPUT_ROOT=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/vbench2_eval/step_100/vbench2_results SELECTED_INFO_JSON=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/vbench2_eval/step_100/vbench2_selected_info.json bash run_vbench2_eval_only.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-2 100" + + jobs: + - name: ywang29-vbench2-eval-yw006-4-token-credit-video-adv-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/submit_vbench2_generate_shards.sh b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_0.yaml b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..067e861544492905bd9229437c2dd5d4c1880127 --- /dev/null +++ b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-dance-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-dance-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_1.yaml b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f6d68b6a571dad2a95968a7e975d892b036df193 --- /dev/null +++ b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-dance-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-dance-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_2.yaml b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..6a3edf5a50691a953a519fdd05abb187a9b73c81 --- /dev/null +++ b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-dance-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-dance-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_3.yaml b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ff583be6ea5ee0e02603f9db19583079d9813cde --- /dev/null +++ b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-dance-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-dance-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_4.yaml b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0565cc3376afd99c1e1c26e691d9857f8641d34c --- /dev/null +++ b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-dance-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-dance-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_5.yaml b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..01e9b940b6cbedcb4a3361efe6f9a9cb844af653 --- /dev/null +++ b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-dance-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-dance-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_6.yaml b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1e13720aec1ac5ace5b9c81ccf125c8349699393 --- /dev/null +++ b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-dance-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-dance-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_7.yaml b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..39cb9d578a02e8b969997a66e1ba386c10b1a100 --- /dev/null +++ b/deploy/vbench2_generate_8_shards_yw006_4_dance_qwen35_9b_token_credit_video_adv/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-dance-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-dance-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_0.yaml b/deploy/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f027eead2058831e01e82a1f20351558c5fb62a3 --- /dev/null +++ b/deploy/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_1.yaml b/deploy/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9993ad36fb873d35b9515958804fa84763ebf236 --- /dev/null +++ b/deploy/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_10.yaml b/deploy/vbench2_generate_shard_10.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9a339e700bc87123fee0fbe31d1484a3e94e6012 --- /dev/null +++ b/deploy/vbench2_generate_shard_10.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=10 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-10 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_11.yaml b/deploy/vbench2_generate_shard_11.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4d366322ffdd5a0ab63e3d64d8051b702b9a9542 --- /dev/null +++ b/deploy/vbench2_generate_shard_11.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=11 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-11 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_12.yaml b/deploy/vbench2_generate_shard_12.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ca0bd4d3b9c6ace658522ee94c472ce257307ec5 --- /dev/null +++ b/deploy/vbench2_generate_shard_12.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=12 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-12 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_13.yaml b/deploy/vbench2_generate_shard_13.yaml new file mode 100644 index 0000000000000000000000000000000000000000..aefed7ccb0567ecd2070395e8579beb53d4723d9 --- /dev/null +++ b/deploy/vbench2_generate_shard_13.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=13 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-13 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_14.yaml b/deploy/vbench2_generate_shard_14.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fdc043403a050132c8d3037f159b4066459fa818 --- /dev/null +++ b/deploy/vbench2_generate_shard_14.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=14 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-14 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_15.yaml b/deploy/vbench2_generate_shard_15.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f7602623b0f6baa3ceb26780e48eefbd5b0c77ad --- /dev/null +++ b/deploy/vbench2_generate_shard_15.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=15 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-15 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_2.yaml b/deploy/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1bbaad271776d28a65569936a627349cf17be8c0 --- /dev/null +++ b/deploy/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_3.yaml b/deploy/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..2294adf7a96a5909e92185e5ce8e1cdfd9103743 --- /dev/null +++ b/deploy/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_4.yaml b/deploy/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c04f4e98ccc201950bb7789698d70b24a1f6cb2f --- /dev/null +++ b/deploy/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_5.yaml b/deploy/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..54a91dd5c486df666f04013b98df5146242be492 --- /dev/null +++ b/deploy/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_6.yaml b/deploy/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1738c60cc0cb8cdb0e544d708745439d9e712f37 --- /dev/null +++ b/deploy/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_7.yaml b/deploy/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..87b822144ed92d08fdea78e62c8b3a1279e14824 --- /dev/null +++ b/deploy/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_8.yaml b/deploy/vbench2_generate_shard_8.yaml new file mode 100644 index 0000000000000000000000000000000000000000..62f1f3d35558e76f618e0787bd2c7536d61ae62c --- /dev/null +++ b/deploy/vbench2_generate_shard_8.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=8 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-8 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/vbench2_generate_shard_9.yaml b/deploy/vbench2_generate_shard_9.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4ebc6b28bb593b332b60b3f5b86a46355101351b --- /dev/null +++ b/deploy/vbench2_generate_shard_9.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=16 PROMPT_SHARD_INDEX=9 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-flow-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-flow-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-x-shard-9 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/base_model/job_name_random_suffix.txt b/deploy/videogen_eval/base_model/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..7042512c3b19bf9515e7c3119d32532c06546490 --- /dev/null +++ b/deploy/videogen_eval/base_model/job_name_random_suffix.txt @@ -0,0 +1 @@ +vg-base-rbed79f diff --git a/deploy/videogen_eval/base_model/submit_videogen_generate_seeds.sh b/deploy/videogen_eval/base_model/submit_videogen_generate_seeds.sh new file mode 100644 index 0000000000000000000000000000000000000000..67c477cd3de004601f31bbd308809545a1d5a7f5 --- /dev/null +++ b/deploy/videogen_eval/base_model/submit_videogen_generate_seeds.sh @@ -0,0 +1,24 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for yaml_path in "${SCRIPT_DIR}"/videogen_generate_seed_*.yaml; do + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/videogen_eval/base_model/videogen_generate_seed_123.yaml b/deploy/videogen_eval/base_model/videogen_generate_seed_123.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5acf1f1574ad1dc7b50c6caaa4cf99074ea355a5 --- /dev/null +++ b/deploy/videogen_eval/base_model/videogen_generate_seed_123.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && RUN_NAME=base_model STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=false SUB_FOLDER=videogen_val1 SEED=123 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-base-rbed79f-s123 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/base_model/videogen_generate_seed_456.yaml b/deploy/videogen_eval/base_model/videogen_generate_seed_456.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d0c47379e3dcc094ab64a8b9b7ce1d806d29edbe --- /dev/null +++ b/deploy/videogen_eval/base_model/videogen_generate_seed_456.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && RUN_NAME=base_model STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=false SUB_FOLDER=videogen_val2 SEED=456 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-base-rbed79f-s456 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/base_model/videogen_generate_seed_789.yaml b/deploy/videogen_eval/base_model/videogen_generate_seed_789.yaml new file mode 100644 index 0000000000000000000000000000000000000000..db198fbeb0fe8b657e6d1b36090cc796807e2bad --- /dev/null +++ b/deploy/videogen_eval/base_model/videogen_generate_seed_789.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && RUN_NAME=base_model STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=false SUB_FOLDER=videogen_val3 SEED=789 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-base-rbed79f-s789 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/submit_all_videogen_generate.sh b/deploy/videogen_eval/submit_all_videogen_generate.sh new file mode 100644 index 0000000000000000000000000000000000000000..767a9f8fe40dae2af1c24fe325aaa0ab4587d6b8 --- /dev/null +++ b/deploy/videogen_eval/submit_all_videogen_generate.sh @@ -0,0 +1,10 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +DRY_RUN="${DRY_RUN:-false}" + +for submit in "${SCRIPT_DIR}"/*/submit_videogen_generate_seeds.sh; do + echo "[run] $submit" + DRY_RUN="$DRY_RUN" bash "$submit" +done diff --git a/deploy/videogen_eval/yw002-1-sage-baseline/job_name_random_suffix.txt b/deploy/videogen_eval/yw002-1-sage-baseline/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..0339634351dd46fe59ad119a69a54e3089dd9a66 --- /dev/null +++ b/deploy/videogen_eval/yw002-1-sage-baseline/job_name_random_suffix.txt @@ -0,0 +1 @@ +vg-1base-r55f10f diff --git a/deploy/videogen_eval/yw002-1-sage-baseline/submit_videogen_generate_seeds.sh b/deploy/videogen_eval/yw002-1-sage-baseline/submit_videogen_generate_seeds.sh new file mode 100644 index 0000000000000000000000000000000000000000..67c477cd3de004601f31bbd308809545a1d5a7f5 --- /dev/null +++ b/deploy/videogen_eval/yw002-1-sage-baseline/submit_videogen_generate_seeds.sh @@ -0,0 +1,24 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for yaml_path in "${SCRIPT_DIR}"/videogen_generate_seed_*.yaml; do + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/videogen_eval/yw002-1-sage-baseline/videogen_generate_seed_123.yaml b/deploy/videogen_eval/yw002-1-sage-baseline/videogen_generate_seed_123.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f5ee787f7138a627d46a23dedcb9e33330a08b46 --- /dev/null +++ b/deploy/videogen_eval/yw002-1-sage-baseline/videogen_generate_seed_123.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw002-1-sage-baseline/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw002-1-sage-baseline/yw002-1-sage-baseline/checkpoint-100/transformer /root/yw002-1-sage-baseline/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw002-1-sage-baseline STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val1 SEED=123 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-1base-r55f10f-s123 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw002-1-sage-baseline/videogen_generate_seed_456.yaml b/deploy/videogen_eval/yw002-1-sage-baseline/videogen_generate_seed_456.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4f18188cb75858e260e9b0d5e7c3624f2ae8d83e --- /dev/null +++ b/deploy/videogen_eval/yw002-1-sage-baseline/videogen_generate_seed_456.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw002-1-sage-baseline/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw002-1-sage-baseline/yw002-1-sage-baseline/checkpoint-100/transformer /root/yw002-1-sage-baseline/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw002-1-sage-baseline STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top100.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val2 SEED=456 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-1base-r55f10f-s456 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw002-1-sage-baseline/videogen_generate_seed_789.yaml b/deploy/videogen_eval/yw002-1-sage-baseline/videogen_generate_seed_789.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0eb2b0483bbdf60d49d02e48bcc9d00663819ebe --- /dev/null +++ b/deploy/videogen_eval/yw002-1-sage-baseline/videogen_generate_seed_789.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw002-1-sage-baseline/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw002-1-sage-baseline/yw002-1-sage-baseline/checkpoint-100/transformer /root/yw002-1-sage-baseline/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw002-1-sage-baseline STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val3 SEED=789 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-1base-r55f10f-s789 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/job_name_random_suffix.txt b/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..6ba5b05dfd169b4e01c5751d26f8856dee038fe7 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/job_name_random_suffix.txt @@ -0,0 +1 @@ +vg-4gemme4b1-rff8d3d diff --git a/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/submit_videogen_generate_seeds.sh b/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/submit_videogen_generate_seeds.sh new file mode 100644 index 0000000000000000000000000000000000000000..67c477cd3de004601f31bbd308809545a1d5a7f5 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/submit_videogen_generate_seeds.sh @@ -0,0 +1,24 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for yaml_path in "${SCRIPT_DIR}"/videogen_generate_seed_*.yaml; do + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/videogen_generate_seed_123.yaml b/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/videogen_generate_seed_123.yaml new file mode 100644 index 0000000000000000000000000000000000000000..960842555d239428932316c07ef42cbb06f760d2 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/videogen_generate_seed_123.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/checkpoint-100/transformer /root/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw006-4-sage-gemma4-e4b-token-credit-video-adv-1 STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val1 SEED=123 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-4gemme4b1-rff8d3d-s123 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/videogen_generate_seed_456.yaml b/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/videogen_generate_seed_456.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fc13d9c49142b1f31cb2eab3a23c64b1af18383b --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/videogen_generate_seed_456.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/checkpoint-100/transformer /root/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw006-4-sage-gemma4-e4b-token-credit-video-adv-1 STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val2 SEED=456 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-4gemme4b1-rff8d3d-s456 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/videogen_generate_seed_789.yaml b/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/videogen_generate_seed_789.yaml new file mode 100644 index 0000000000000000000000000000000000000000..68f4ff5dfa92bd2c344865d8155df9273079e414 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/videogen_generate_seed_789.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/checkpoint-100/transformer /root/yw006-4-sage-gemma4-e4b-token-credit-video-adv-1/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw006-4-sage-gemma4-e4b-token-credit-video-adv-1 STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val3 SEED=789 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-4gemme4b1-rff8d3d-s789 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/job_name_random_suffix.txt b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..db54eae16eab433080f9e07b0b40f582315c134a --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/job_name_random_suffix.txt @@ -0,0 +1 @@ +vg-42-rffc5f8 diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/submit_videogen_generate_seeds.sh b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/submit_videogen_generate_seeds.sh new file mode 100644 index 0000000000000000000000000000000000000000..67c477cd3de004601f31bbd308809545a1d5a7f5 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/submit_videogen_generate_seeds.sh @@ -0,0 +1,24 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for yaml_path in "${SCRIPT_DIR}"/videogen_generate_seed_*.yaml; do + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/videogen_generate_seed_123.yaml b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/videogen_generate_seed_123.yaml new file mode 100644 index 0000000000000000000000000000000000000000..134a506f52953d2b001682ee2341bed46e1b6bfc --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/videogen_generate_seed_123.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/checkpoint-100/transformer /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw006-4-sage-qwen35-9b-token-credit-video-adv-2 STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val1 SEED=123 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-42-rffc5f8-s123 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/videogen_generate_seed_456.yaml b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/videogen_generate_seed_456.yaml new file mode 100644 index 0000000000000000000000000000000000000000..90f145c5d67f5b122ac9eb36c07d43ec84727760 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/videogen_generate_seed_456.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/checkpoint-100/transformer /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw006-4-sage-qwen35-9b-token-credit-video-adv-2 STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val2 SEED=456 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-42-rffc5f8-s456 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/videogen_generate_seed_789.yaml b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/videogen_generate_seed_789.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fa7c52ac72c54f487cc9f6dc1c7a8899b934e655 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/videogen_generate_seed_789.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/checkpoint-100/transformer /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw006-4-sage-qwen35-9b-token-credit-video-adv-2 STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val3 SEED=789 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-42-rffc5f8-s789 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/job_name_random_suffix.txt b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..b303b95fdaea7836f4faeea7a058de56b574ee26 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/job_name_random_suffix.txt @@ -0,0 +1 @@ +vg-4framabl3-rca6077 diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/submit_videogen_generate_seeds.sh b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/submit_videogen_generate_seeds.sh new file mode 100644 index 0000000000000000000000000000000000000000..67c477cd3de004601f31bbd308809545a1d5a7f5 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/submit_videogen_generate_seeds.sh @@ -0,0 +1,24 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for yaml_path in "${SCRIPT_DIR}"/videogen_generate_seed_*.yaml; do + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/videogen_generate_seed_123.yaml b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/videogen_generate_seed_123.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b48e1aa0f1dd4104e2d9b2ebe208295ab2861e85 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/videogen_generate_seed_123.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/checkpoint-100/transformer /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30 STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val1 SEED=123 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-4framabl3-rca6077-s123 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/videogen_generate_seed_456.yaml b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/videogen_generate_seed_456.yaml new file mode 100644 index 0000000000000000000000000000000000000000..65390b9b595b51ea1a1273e05b6e10544dee4f26 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/videogen_generate_seed_456.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/checkpoint-100/transformer /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30 STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val2 SEED=456 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-4framabl3-rca6077-s456 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/videogen_generate_seed_789.yaml b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/videogen_generate_seed_789.yaml new file mode 100644 index 0000000000000000000000000000000000000000..125d7878bcd6d0a4a147c6521e7ad4ef40e40633 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/videogen_generate_seed_789.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/checkpoint-100/transformer /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw006-4-sage-qwen35-9b-token-credit-video-adv-frames-abl30 STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val3 SEED=789 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-4framabl3-rca6077-s789 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/job_name_random_suffix.txt b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..ead66fb5245c04e4ff1fa469e3858e1f3561d580 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/job_name_random_suffix.txt @@ -0,0 +1 @@ +vg-4rati-r2f77a9 diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/submit_videogen_generate_seeds.sh b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/submit_videogen_generate_seeds.sh new file mode 100644 index 0000000000000000000000000000000000000000..67c477cd3de004601f31bbd308809545a1d5a7f5 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/submit_videogen_generate_seeds.sh @@ -0,0 +1,24 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for yaml_path in "${SCRIPT_DIR}"/videogen_generate_seed_*.yaml; do + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/videogen_generate_seed_123.yaml b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/videogen_generate_seed_123.yaml new file mode 100644 index 0000000000000000000000000000000000000000..32b959f971271ec7a358e9f75f997ac828041be7 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/videogen_generate_seed_123.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/checkpoint-100/transformer /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5 STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val1 SEED=123 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-4rati-r2f77a9-s123 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/videogen_generate_seed_456.yaml b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/videogen_generate_seed_456.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d225f74bff158a3b8af59f7d92f86a5613232219 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/videogen_generate_seed_456.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/checkpoint-100/transformer /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5 STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val2 SEED=456 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-4rati-r2f77a9-s456 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/videogen_generate_seed_789.yaml b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/videogen_generate_seed_789.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7fb94682abca74a8ffbbeeda924bdb99a1734f32 --- /dev/null +++ b/deploy/videogen_eval/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/videogen_generate_seed_789.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/checkpoint-100/transformer /root/yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw006-4-sage-qwen35-9b-token-credit-video-adv-rating5 STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val3 SEED=789 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-4rati-r2f77a9-s789 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/job_name_random_suffix.txt b/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..95778cf36ab5363107b4d5e572c0e78bcf8261e2 --- /dev/null +++ b/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/job_name_random_suffix.txt @@ -0,0 +1 @@ +vg-1vide-r3deee5 diff --git a/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/submit_videogen_generate_seeds.sh b/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/submit_videogen_generate_seeds.sh new file mode 100644 index 0000000000000000000000000000000000000000..67c477cd3de004601f31bbd308809545a1d5a7f5 --- /dev/null +++ b/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/submit_videogen_generate_seeds.sh @@ -0,0 +1,24 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for yaml_path in "${SCRIPT_DIR}"/videogen_generate_seed_*.yaml; do + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/videogen_generate_seed_123.yaml b/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/videogen_generate_seed_123.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f8c16f02ae92fbd94f8619f3e273c4920a6b6e82 --- /dev/null +++ b/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/videogen_generate_seed_123.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw007-1-sage-videoalign-token-credit-video-adv/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw007-1-sage-videoalign-token-credit-video-adv/yw007-1-sage-videoalign-token-credit-video-adv/checkpoint-100/transformer /root/yw007-1-sage-videoalign-token-credit-video-adv/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw007-1-sage-videoalign-token-credit-video-adv STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val1 SEED=123 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-1vide-r3deee5-s123 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/videogen_generate_seed_456.yaml b/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/videogen_generate_seed_456.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0f4d3b580abd777fa1e4e4d41687754a6637f671 --- /dev/null +++ b/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/videogen_generate_seed_456.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw007-1-sage-videoalign-token-credit-video-adv/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw007-1-sage-videoalign-token-credit-video-adv/yw007-1-sage-videoalign-token-credit-video-adv/checkpoint-100/transformer /root/yw007-1-sage-videoalign-token-credit-video-adv/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw007-1-sage-videoalign-token-credit-video-adv STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val2 SEED=456 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-1vide-r3deee5-s456 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/videogen_generate_seed_789.yaml b/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/videogen_generate_seed_789.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a4098961e1b8ef38ee95a50800ccaaccad5514d6 --- /dev/null +++ b/deploy/videogen_eval/yw007-1-sage-videoalign-token-credit-video-adv/videogen_generate_seed_789.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw007-1-sage-videoalign-token-credit-video-adv/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw007-1-sage-videoalign-token-credit-video-adv/yw007-1-sage-videoalign-token-credit-video-adv/checkpoint-100/transformer /root/yw007-1-sage-videoalign-token-credit-video-adv/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw007-1-sage-videoalign-token-credit-video-adv STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val3 SEED=789 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-1vide-r3deee5-s789 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/job_name_random_suffix.txt b/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..37e5f5bba485e39f1291e11877523ac0044399b9 --- /dev/null +++ b/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/job_name_random_suffix.txt @@ -0,0 +1 @@ +vg-1vide-rc8956a diff --git a/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/submit_videogen_generate_seeds.sh b/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/submit_videogen_generate_seeds.sh new file mode 100644 index 0000000000000000000000000000000000000000..67c477cd3de004601f31bbd308809545a1d5a7f5 --- /dev/null +++ b/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/submit_videogen_generate_seeds.sh @@ -0,0 +1,24 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for yaml_path in "${SCRIPT_DIR}"/videogen_generate_seed_*.yaml; do + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/videogen_generate_seed_123.yaml b/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/videogen_generate_seed_123.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0f97642003fe468a3d35f4247b8b029d797cd313 --- /dev/null +++ b/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/videogen_generate_seed_123.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw008-1-sage-videoscore2-token-credit-video-adv/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw008-1-sage-videoscore2-token-credit-video-adv/yw008-1-sage-videoscore2-token-credit-video-adv/checkpoint-100/transformer /root/yw008-1-sage-videoscore2-token-credit-video-adv/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw008-1-sage-videoscore2-token-credit-video-adv STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val1 SEED=123 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-1vide-rc8956a-s123 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/videogen_generate_seed_456.yaml b/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/videogen_generate_seed_456.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f599633b948066f754b1407787cd0d920a055639 --- /dev/null +++ b/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/videogen_generate_seed_456.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw008-1-sage-videoscore2-token-credit-video-adv/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw008-1-sage-videoscore2-token-credit-video-adv/yw008-1-sage-videoscore2-token-credit-video-adv/checkpoint-100/transformer /root/yw008-1-sage-videoscore2-token-credit-video-adv/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw008-1-sage-videoscore2-token-credit-video-adv STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val2 SEED=456 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-1vide-rc8956a-s456 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/videogen_generate_seed_789.yaml b/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/videogen_generate_seed_789.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5a7991f162de3a49f5c3ef06ac574f474943c5a5 --- /dev/null +++ b/deploy/videogen_eval/yw008-1-sage-videoscore2-token-credit-video-adv/videogen_generate_seed_789.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && mkdir -p /root/yw008-1-sage-videoscore2-token-credit-video-adv/checkpoint-100/transformer && AWS_PROFILE=default_mle aws s3 sync s3://snap-genvid-us-east-2/ywang29/checkpoints_large_model/yw008-1-sage-videoscore2-token-credit-video-adv/yw008-1-sage-videoscore2-token-credit-video-adv/checkpoint-100/transformer /root/yw008-1-sage-videoscore2-token-credit-video-adv/checkpoint-100/transformer --only-show-errors && RUN_NAME=yw008-1-sage-videoscore2-token-credit-video-adv STEP=100 VALID_VIDEO_CSV=assets/video_gen_eval_top400.csv USE_CHECKPOINT=true SUB_FOLDER=videogen_val3 SEED=789 N_INFERENCE_GPU=8 bash generate.sh" + + jobs: + - name: ywang29-videogen-vg-1vide-rc8956a-s789 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw003-1-dance-baseline-step_200/job_name_random_suffix.txt b/deploy/yw003-1-dance-baseline-step_200/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..668c73f7918b1080e69b0f1ccf74dfd9e30bf3aa --- /dev/null +++ b/deploy/yw003-1-dance-baseline-step_200/job_name_random_suffix.txt @@ -0,0 +1 @@ +re92359 diff --git a/deploy/yw003-1-dance-baseline-step_200/submit_vbench2_generate_shards.sh b/deploy/yw003-1-dance-baseline-step_200/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw003-1-dance-baseline-step_200/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_0.yaml b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..44481313936e20431d82817ccadd971eb9a23674 --- /dev/null +++ b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw003-1-dance-baseline/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw003-1-dance-baseline 200" + + jobs: + - name: ywang29-vbench2-gen-dance200-re92359-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_1.yaml b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9e435ec0f136b09669c4fe3097f8f4fdc29e61a1 --- /dev/null +++ b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw003-1-dance-baseline/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw003-1-dance-baseline 200" + + jobs: + - name: ywang29-vbench2-gen-dance200-re92359-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_2.yaml b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d2db8d8a4f4e5d1b08611223d26598a3e1e2d24f --- /dev/null +++ b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw003-1-dance-baseline/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw003-1-dance-baseline 200" + + jobs: + - name: ywang29-vbench2-gen-dance200-re92359-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_3.yaml b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..60e13f32f3899ba37cd54e413f76c185888c9189 --- /dev/null +++ b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw003-1-dance-baseline/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw003-1-dance-baseline 200" + + jobs: + - name: ywang29-vbench2-gen-dance200-re92359-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_4.yaml b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..db908c854a5702d1f82b3c052cc473c77838ef24 --- /dev/null +++ b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw003-1-dance-baseline/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw003-1-dance-baseline 200" + + jobs: + - name: ywang29-vbench2-gen-dance200-re92359-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_5.yaml b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4fb14c28661b33def1a937a11a1f766a992aa08f --- /dev/null +++ b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw003-1-dance-baseline/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw003-1-dance-baseline 200" + + jobs: + - name: ywang29-vbench2-gen-dance200-re92359-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_6.yaml b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..2bf92c329b37a7c333e54d0695baa487a526e554 --- /dev/null +++ b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw003-1-dance-baseline/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw003-1-dance-baseline 200" + + jobs: + - name: ywang29-vbench2-gen-dance200-re92359-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_7.yaml b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..884e83e2587af12f3e5c7e0f984b3538486e50e2 --- /dev/null +++ b/deploy/yw003-1-dance-baseline-step_200/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw003-1-dance-baseline/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw003-1-dance-baseline 200" + + jobs: + - name: ywang29-vbench2-gen-dance200-re92359-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw005-cps-baseline/job_name_random_suffix.txt b/deploy/yw005-cps-baseline/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..4e74b3404e58664af4c8a0b474520267fac7c5c3 --- /dev/null +++ b/deploy/yw005-cps-baseline/job_name_random_suffix.txt @@ -0,0 +1 @@ +rfcc1f8 diff --git a/deploy/yw005-cps-baseline/submit_vbench2_generate_shards.sh b/deploy/yw005-cps-baseline/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw005-cps-baseline/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw005-cps-baseline/vbench2_generate_shard_0.yaml b/deploy/yw005-cps-baseline/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b56adc21a90d3691376629e2bd05dc277d2091a5 --- /dev/null +++ b/deploy/yw005-cps-baseline/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw005-cps-baseline/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw005-cps-baseline 100" + + jobs: + - name: ywang29-vbench2-gen-cpsbase-rfcc1f8-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw005-cps-baseline/vbench2_generate_shard_1.yaml b/deploy/yw005-cps-baseline/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f4157eabe91caae8c08746c33c3dbc74bcb4f8b3 --- /dev/null +++ b/deploy/yw005-cps-baseline/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw005-cps-baseline/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw005-cps-baseline 100" + + jobs: + - name: ywang29-vbench2-gen-cpsbase-rfcc1f8-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw005-cps-baseline/vbench2_generate_shard_2.yaml b/deploy/yw005-cps-baseline/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..798232923e610ef3aa36aef75f90bcae6e2b7289 --- /dev/null +++ b/deploy/yw005-cps-baseline/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw005-cps-baseline/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw005-cps-baseline 100" + + jobs: + - name: ywang29-vbench2-gen-cpsbase-rfcc1f8-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw005-cps-baseline/vbench2_generate_shard_3.yaml b/deploy/yw005-cps-baseline/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..6e9ce1d9e9e1fb67a329c8c48649f088785dec0b --- /dev/null +++ b/deploy/yw005-cps-baseline/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw005-cps-baseline/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw005-cps-baseline 100" + + jobs: + - name: ywang29-vbench2-gen-cpsbase-rfcc1f8-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw005-cps-baseline/vbench2_generate_shard_4.yaml b/deploy/yw005-cps-baseline/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5fb4723490c3a2f771f6d60b15722267887ceddc --- /dev/null +++ b/deploy/yw005-cps-baseline/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw005-cps-baseline/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw005-cps-baseline 100" + + jobs: + - name: ywang29-vbench2-gen-cpsbase-rfcc1f8-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw005-cps-baseline/vbench2_generate_shard_5.yaml b/deploy/yw005-cps-baseline/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d5166d86ff8b88437cce1f390a53e9b038bdc2d6 --- /dev/null +++ b/deploy/yw005-cps-baseline/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw005-cps-baseline/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw005-cps-baseline 100" + + jobs: + - name: ywang29-vbench2-gen-cpsbase-rfcc1f8-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw005-cps-baseline/vbench2_generate_shard_6.yaml b/deploy/yw005-cps-baseline/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..2e76c450fc976d1b09c8d31267b54d6c7883dbbd --- /dev/null +++ b/deploy/yw005-cps-baseline/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw005-cps-baseline/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw005-cps-baseline 100" + + jobs: + - name: ywang29-vbench2-gen-cpsbase-rfcc1f8-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw005-cps-baseline/vbench2_generate_shard_7.yaml b/deploy/yw005-cps-baseline/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a5ef1382be7376eee16a7c94691cc2d6b2653736 --- /dev/null +++ b/deploy/yw005-cps-baseline/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw005-cps-baseline/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw005-cps-baseline 100" + + jobs: + - name: ywang29-vbench2-gen-cpsbase-rfcc1f8-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-2-sage-qwen35-9b-free-generation/submit_vbench2_generate_shards.sh b/deploy/yw006-2-sage-qwen35-9b-free-generation/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-2-sage-qwen35-9b-free-generation/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_0.yaml b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..10032383c27cd0e6e68c46de4103117d9b975f66 --- /dev/null +++ b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-2-sage-qwen35-9b-free-generation/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-2-sage-qwen35-9b-free-generation 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_1.yaml b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..742266ec9d4980c9f8ac03ca113924e572d2c63c --- /dev/null +++ b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-2-sage-qwen35-9b-free-generation/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-2-sage-qwen35-9b-free-generation 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_2.yaml b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..22258726665d67411e137067e1d0505319136114 --- /dev/null +++ b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-2-sage-qwen35-9b-free-generation/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-2-sage-qwen35-9b-free-generation 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_3.yaml b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8e3c9fd422a2a507a3463f7e14a350bd30603a19 --- /dev/null +++ b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-2-sage-qwen35-9b-free-generation/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-2-sage-qwen35-9b-free-generation 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_4.yaml b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8d445162124eef42406df0bb73f34824d3549b33 --- /dev/null +++ b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-2-sage-qwen35-9b-free-generation/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-2-sage-qwen35-9b-free-generation 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_5.yaml b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..024caaba2edb03da3c9655c82b5c9b1d6c8a0e05 --- /dev/null +++ b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-2-sage-qwen35-9b-free-generation/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-2-sage-qwen35-9b-free-generation 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_6.yaml b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..993f53f25a615284b577f916a31132a3f7990453 --- /dev/null +++ b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-2-sage-qwen35-9b-free-generation/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-2-sage-qwen35-9b-free-generation 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_7.yaml b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1faf8aea0931b5d64cded32eecb51440cc8c480d --- /dev/null +++ b/deploy/yw006-2-sage-qwen35-9b-free-generation/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-2-sage-qwen35-9b-free-generation/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-2-sage-qwen35-9b-free-generation 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/job_name_random_suffix.txt b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..df4ca81ec47ff4e86c4c3e1efb332cf2d11c2269 --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/job_name_random_suffix.txt @@ -0,0 +1 @@ +rfe2858 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/submit_vbench2_generate_shards.sh b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_0.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..23c00695b50ddba6edd46ada17614a296cae837a --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-4b42v1-rfe2858-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_1.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..3ad7df15880ea20a125bf14052b3b1c0369aa17f --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-4b42v1-rfe2858-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_2.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0ac280f90d5f4686c83ee8a0a6c74895f3b74ef5 --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-4b42v1-rfe2858-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_3.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1d7aae34b64ae02a8b775e20c7b0df31c66c5745 --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-4b42v1-rfe2858-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_4.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4f28df0105962beab3d3a5bcfb1424200aedeb96 --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-4b42v1-rfe2858-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_5.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..429b1cba9d74e5026a1caba3adf011fe3a0ee2af --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-4b42v1-rfe2858-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_6.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..74f945377d91087ca38feb5ab45aadfbc6ba9529 --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-4b42v1-rfe2858-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_7.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1513fd3eff784d10f682499540062a4c9360c27f --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv-1 100" + + jobs: + - name: ywang29-vbench2-gen-4b42v1-rfe2858-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/submit_vbench2_generate_shards.sh b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_0.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fe6cf651b4ae5cf5aae88ee6de78c90fb72bd4f5 --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-4b42-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_1.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ef6215efeb4f594e3891a1149e70f1313d748748 --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-4b42-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_2.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b6ee92ed88f160fa55255c4e1a1b59f167152cfd --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-4b42-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_3.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bfc56ac39779968e88a113a4ce7d55ded4115043 --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-4b42-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_4.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b5b19e29dc56ff08c483ad901c0539a1c1a39766 --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-4b42-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_5.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5e8b0e5270c84d2e239a01b3cfe2ac6186a6ca04 --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-4b42-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_6.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..28dcbcd1050a0d810d999f189fd447c0248fe328 --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-4b42-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_7.yaml b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f723be31cc563256d39c9317ab062f37affab4bd --- /dev/null +++ b/deploy/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-2-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-2-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-4b42-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/job_name_random_suffix.txt b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..546800f041505c4811c1476ac73cd463358a6e1d --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/job_name_random_suffix.txt @@ -0,0 +1 @@ +r981477 diff --git a/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/submit_vbench2_generate_shards.sh b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_0.yaml b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b73715113d3d446fadbe4c82009b36279bc2f950 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-08b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-08b44-r981477-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_1.yaml b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7961484e195bcc1c7aad51662faa777611a9433c --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-08b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-08b44-r981477-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_2.yaml b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0f49fe16526cdf305822f199ec354221eaf951a3 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-08b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-08b44-r981477-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_3.yaml b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bc402f7106bcb41ddec671ad51c15aa5cb652f74 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-08b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-08b44-r981477-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_4.yaml b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a78ee413e5d6300a6051e7c446734de03ad2bc19 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-08b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-08b44-r981477-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_5.yaml b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ab738225af81d61451a14ee3dc0ca61bd7471ec3 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-08b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-08b44-r981477-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_6.yaml b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0b9c38ce3592f8142f47c9490e82f04fe9675d19 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-08b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-08b44-r981477-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_7.yaml b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7d237a0ed7c9a7355a56bc3387eac050a7835698 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-08b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-08b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-08b44-r981477-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/submit_vbench2_generate_shards.sh b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_0.yaml b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..66eab219c694311855283a386e8659eff866bff4 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-y-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_1.yaml b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4b325f6f35548fe07a31867ffe3f04f5516ce54a --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-y-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_2.yaml b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..2e6363bd5f7cb2cedb53899cd0bad6eb914c11f7 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-y-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_3.yaml b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..186b04b096e72c28ddb6e305b9ce2e07606b7001 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-y-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_4.yaml b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..81d31d16932651ec1f3b7550dac968f58d960c26 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-y-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_5.yaml b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8c92ae66dbd578cf230472f0a86ce0d54a2e319c --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-y-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_6.yaml b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1c31d6b9cb8df7e630e44b511620075bda141c9c --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-y-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_7.yaml b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..3990f56db4b2bd42354be4538623143cad4211e1 --- /dev/null +++ b/deploy/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-4-sage-qwen35-4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-4-sage-qwen35-4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-y-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/job_name_random_suffix.txt b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..47f2f5976981fe7d9ce2dd2abc4c9ef0cc4ebffd --- /dev/null +++ b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/job_name_random_suffix.txt @@ -0,0 +1 @@ +r781740 diff --git a/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/submit_vbench2_generate_shards.sh b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_0.yaml b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c632c863fd96200591ad77e80233f089c09b4bfa --- /dev/null +++ b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-cpsva-r781740-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_1.yaml b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a865e77ccfb926ef596fdaa96f14eb5d248410e4 --- /dev/null +++ b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-cpsva-r781740-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_2.yaml b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1b92be3be802987fd06cdd466e26911ebc98a3f0 --- /dev/null +++ b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-cpsva-r781740-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_3.yaml b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0758a301f6665013c632714801fbdd8bb358f3a7 --- /dev/null +++ b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-cpsva-r781740-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_4.yaml b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..66213c8a20c8e1933b51dbc281594963e3922a9b --- /dev/null +++ b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-cpsva-r781740-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_5.yaml b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..cdad28f667645fb7f326bbbf3f37d324caa94255 --- /dev/null +++ b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-cpsva-r781740-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_6.yaml b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d542884416deec246ac8adbfad20189ff381bf6f --- /dev/null +++ b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-cpsva-r781740-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_7.yaml b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f5515f8d75de047fafa9c8e4ab1119027e901f26 --- /dev/null +++ b/deploy/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-cps-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-cps-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-cpsva-r781740-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/job_name_random_suffix.txt b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..170d8d840bf69e36cb18ed3be4d7462775e130ac --- /dev/null +++ b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/job_name_random_suffix.txt @@ -0,0 +1 @@ +rdae5ef diff --git a/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/submit_vbench2_generate_shards.sh b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_0.yaml b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..002017304df6481e5754046dbf5aebd8a105ccca --- /dev/null +++ b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b-rdae5ef-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_1.yaml b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1c29b8443897ebc2f67816f054c42ccdaeee316e --- /dev/null +++ b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b-rdae5ef-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_2.yaml b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ab78f2ee2e67f4702a6803b4f6c5d45ea2331347 --- /dev/null +++ b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b-rdae5ef-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_3.yaml b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..509da1e0b1e95fc1312daaad8612ab1bc3d82179 --- /dev/null +++ b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b-rdae5ef-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_4.yaml b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..94eb6cf8cb6e61cfccfd4506aaa41f0f6b90118c --- /dev/null +++ b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b-rdae5ef-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_5.yaml b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1c0256c226d0b5719f76a7dbce4287f77582cc98 --- /dev/null +++ b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b-rdae5ef-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_6.yaml b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4444d20e52dbad694661f6118e11094ec55f30a3 --- /dev/null +++ b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b-rdae5ef-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_7.yaml b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..400c654a4f5eac65bfb455b24654acd35e2d4c07 --- /dev/null +++ b/deploy/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-gemma4-e4b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-gemma4-e4b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-g4e4b-rdae5ef-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/job_name_random_suffix.txt b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..d74834babe09f3d7247b94098968627d923b1f4b --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/job_name_random_suffix.txt @@ -0,0 +1 @@ +reb51d7 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/submit_vbench2_generate_shards.sh b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_0.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..2c0ff11bc5d203fa51505d3b135c7ee5717b7833 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-2 200" + + jobs: + - name: ywang29-vbench2-gen-adv2s200-reb51d7-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_1.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4b5546bdf56e77e8ee0f5000aeb3798e27002d2b --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-2 200" + + jobs: + - name: ywang29-vbench2-gen-adv2s200-reb51d7-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_2.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c69b25437f23249a4933e0ab5b3daeb204f7ba38 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-2 200" + + jobs: + - name: ywang29-vbench2-gen-adv2s200-reb51d7-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_3.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..70c32243de84c66c8e69b0d4d3cda9a76924faff --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-2 200" + + jobs: + - name: ywang29-vbench2-gen-adv2s200-reb51d7-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_4.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..02a9f19a7c615f64988d456a6505c7417a97fecd --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-2 200" + + jobs: + - name: ywang29-vbench2-gen-adv2s200-reb51d7-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_5.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ba5872185a727c6c9a47477c76c71798a132347b --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-2 200" + + jobs: + - name: ywang29-vbench2-gen-adv2s200-reb51d7-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_6.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8db4ed83f0a606e3b8d374d90519322c4eebd1fb --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-2 200" + + jobs: + - name: ywang29-vbench2-gen-adv2s200-reb51d7-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_7.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f91c189855928e4547585717a7f98a0a8e660af3 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-2-step_200/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-2/vbench2_eval/step_200 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-2 200" + + jobs: + - name: ywang29-vbench2-gen-adv2s200-reb51d7-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/job_name_random_suffix.txt b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..5f0709b489864ac5350160513f426abd913780e2 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/job_name_random_suffix.txt @@ -0,0 +1 @@ +r817118 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/submit_vbench2_generate_shards.sh b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_0.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f4b4721de4e0f177268d8955441914877fa3faaa --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_50 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 50" + + jobs: + - name: ywang29-vbench2-gen-adv3s50-r817118-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_1.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..030daeb7e15d7c99c4c5b6681637c477dea3c1a7 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_50 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 50" + + jobs: + - name: ywang29-vbench2-gen-adv3s50-r817118-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_2.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d798d8124dca2236c169235b5553dab785c7cc84 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_50 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 50" + + jobs: + - name: ywang29-vbench2-gen-adv3s50-r817118-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_3.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..77b40e361f6cebdc02c40633f0837fa6c79dc8e1 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_50 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 50" + + jobs: + - name: ywang29-vbench2-gen-adv3s50-r817118-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_4.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f7765b65c3627c42c2465fd9beab43beaf8f41dc --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_50 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 50" + + jobs: + - name: ywang29-vbench2-gen-adv3s50-r817118-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_5.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7c61d8e085bb53c7707c27630afb524d0c82a9e8 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_50 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 50" + + jobs: + - name: ywang29-vbench2-gen-adv3s50-r817118-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_6.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..551f37e76d085bfddcf7a2955292ed479cb6fe4f --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_50 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 50" + + jobs: + - name: ywang29-vbench2-gen-adv3s50-r817118-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_7.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..deaa745e121b41835af5aae38fff9fe571d90ad6 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3-step_50/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_50 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 50" + + jobs: + - name: ywang29-vbench2-gen-adv3s50-r817118-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/job_name_random_suffix.txt b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..09ed9c81fa365ef560e586de3328a62defd56153 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/job_name_random_suffix.txt @@ -0,0 +1 @@ +r7afc8e diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/submit_vbench2_generate_shards.sh b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_0.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5e69c1cb130e4a06f8204f05b050a902a83ca918 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 100" + + jobs: + - name: ywang29-vbench2-gen-tcva3-r7afc8e-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_1.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..01903cec39aa47a50669ca68e95f7dc792e61a3d --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 100" + + jobs: + - name: ywang29-vbench2-gen-tcva3-r7afc8e-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_2.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..820d9ff368a435a6c4315f125abbded666c3f1cf --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 100" + + jobs: + - name: ywang29-vbench2-gen-tcva3-r7afc8e-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_3.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..08066e3dbeeaf968b174c38631d2953d82ac95a1 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 100" + + jobs: + - name: ywang29-vbench2-gen-tcva3-r7afc8e-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_4.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..95b8bb78dab0ab53dbca781b3b93338c47e149c5 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 100" + + jobs: + - name: ywang29-vbench2-gen-tcva3-r7afc8e-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_5.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..6923ac2c17d154b6087d2ca9cfd8d75f0d506448 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 100" + + jobs: + - name: ywang29-vbench2-gen-tcva3-r7afc8e-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_6.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..41a0f09e6919d03c6c6fa29fd42a4fe85028bc4b --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 100" + + jobs: + - name: ywang29-vbench2-gen-tcva3-r7afc8e-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_7.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..441a2830a690aa849305bfb019620410097d81b3 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-3 100" + + jobs: + - name: ywang29-vbench2-gen-tcva3-r7afc8e-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/submit_vbench2_generate_shards.sh b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_0.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b04409b16cdb1ff5fe265c67c75115cbf8281860 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3 100" + + jobs: + - name: ywang29-vbench2-gen-nw3-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_1.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4bfe9f5151c7aed6efdb53c4ea33ccfa90d3822f --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3 100" + + jobs: + - name: ywang29-vbench2-gen-nw3-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_2.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..baf095362ade6e1fde191b2d40594a7cce12dd00 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3 100" + + jobs: + - name: ywang29-vbench2-gen-nw3-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_3.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..40ac25279b67d9f9703de7af1f5c71174a6f171c --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3 100" + + jobs: + - name: ywang29-vbench2-gen-nw3-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_4.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c64d9b91c9e2750dfa66258607795d5f8120048f --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3 100" + + jobs: + - name: ywang29-vbench2-gen-nw3-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_5.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..87cd73ae0a433e99529477d44e211094a9c8efb2 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3 100" + + jobs: + - name: ywang29-vbench2-gen-nw3-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_6.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d1b37c9b423464f7abfa8d34ea7457828cb143b0 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3 100" + + jobs: + - name: ywang29-vbench2-gen-nw3-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_7.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..09beb64df63fa6538cadd77e8147c041e62ff3ae --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-no-weighting-3 100" + + jobs: + - name: ywang29-vbench2-gen-nw3-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/job_name_random_suffix.txt b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..7d51edcdd6eb5d978c641f5477c134efd3387676 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/job_name_random_suffix.txt @@ -0,0 +1 @@ +r7c5980 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/submit_vbench2_generate_shards.sh b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_0.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4b5ae5f872ad4b004e33181f53c607758c11af8a --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q1abl-r7c5980-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_1.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..be8318b9127b89344677c8d2a965e1bfdcee1e1e --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q1abl-r7c5980-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_2.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5ecdc486a01ca4acd4de58882c1c2ac4bfaf5d1e --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q1abl-r7c5980-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_3.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..86215473149448ccd744ae5e81c4967ea7054486 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q1abl-r7c5980-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_4.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d426437360b1a653a3f4c0f9c0caced882c3faba --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q1abl-r7c5980-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_5.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0b374e8ffe7a1373d1d3c3468063dea3b5c13aab --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q1abl-r7c5980-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_6.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..948e10f984964c2459bf840c58336f22f2110cad --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q1abl-r7c5980-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_7.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b9ffdfa2eadf61b8d315debee930cbdb251d6da4 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question1-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q1abl-r7c5980-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/job_name_random_suffix.txt b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..885ef58ead553c7135410a78c7fe30a05b96fb68 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/job_name_random_suffix.txt @@ -0,0 +1 @@ +r23e461 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/submit_vbench2_generate_shards.sh b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_0.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9b7fba90fbdebbe311cb2c6bfa3e71a85bb41c5f --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q3abl-r23e461-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_1.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..cc8b375a10faf6bb61e16cf2e9bd156dc613b4ab --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q3abl-r23e461-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_2.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b1ee8bfadd1fe3b751f376ea4af38b750ee1fd38 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q3abl-r23e461-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_3.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c4891f2308079bcdc0751c21639b7f86147f77f1 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q3abl-r23e461-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_4.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..79ecbd22016b6dd87bbd96940cfbaab90cd3bb0b --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q3abl-r23e461-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_5.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5b9d74913ef19491f1b363ebdb55c5bd5af14709 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q3abl-r23e461-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_6.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9b81eecc258d52071ac33ee85b5300c248556e99 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q3abl-r23e461-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_7.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5ff0f3f33d6e0081e6e07d0e5cbc3eacb128ea43 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-question3-abl 100" + + jobs: + - name: ywang29-vbench2-gen-q3abl-r23e461-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/submit_vbench2_generate_shards.sh b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_0.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..3c5992a27b19d268965d456f049bd93ccddc0ae8 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2 100" + + jobs: + - name: ywang29-vbench2-gen-sw2-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_1.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..47ebd105b18c7176837ca765c14b8a261f703c5b --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2 100" + + jobs: + - name: ywang29-vbench2-gen-sw2-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_2.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..44cf19830f4e8358b1ae74dc4b88af85b81375fb --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2 100" + + jobs: + - name: ywang29-vbench2-gen-sw2-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_3.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d31b21db73fac2732c85306447e22f46631505d0 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2 100" + + jobs: + - name: ywang29-vbench2-gen-sw2-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_4.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..2cf5eb9dcd1716012c4b614e849ee0fdf10804f0 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2 100" + + jobs: + - name: ywang29-vbench2-gen-sw2-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_5.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8c33c87242ba8b3b06c105b27ac303723cf213f9 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2 100" + + jobs: + - name: ywang29-vbench2-gen-sw2-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_6.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b17a06fce8ca22e6d129f0350d3eb3571b518d84 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2 100" + + jobs: + - name: ywang29-vbench2-gen-sw2-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_7.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ff2d3dc3359ba1e4887fefaf7ad316662b486503 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-2 100" + + jobs: + - name: ywang29-vbench2-gen-sw2-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/job_name_random_suffix.txt b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..13d0aefe87703ebf4d88f2f2f919f495db8a5555 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/job_name_random_suffix.txt @@ -0,0 +1 @@ +r5ec470 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/submit_vbench2_generate_shards.sh b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_0.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1907bc590a581e227bb265623c912d99d7805c02 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcvasw4-r5ec470-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_1.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c1e3f9eda7e20f36221800baa6b84bfb2eaba7e3 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcvasw4-r5ec470-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_2.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..62d621b0cec2500f0f591ea310c8e3f6ae490453 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcvasw4-r5ec470-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_3.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a82eb840ba5737acfe396b4a4ec2c30d68f7cbde --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcvasw4-r5ec470-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_4.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..3b11a1c2488c385cb6f6a212b04f887f80912589 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcvasw4-r5ec470-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_5.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d507ac8055f15baf5ecb2745704f0c7dcc9b3554 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcvasw4-r5ec470-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_6.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..770bf8f1c309bd7c43f3b4eae9cee059f9fbae56 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcvasw4-r5ec470-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_7.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8243a1af5a453756d776a24332ad9b8454742bfa --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-4-sage-qwen35-9b-token-credit-video-adv-shuffled-weighting-4 100" + + jobs: + - name: ywang29-vbench2-gen-tcvasw4-r5ec470-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/submit.sh b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/submit.sh new file mode 100644 index 0000000000000000000000000000000000000000..dcf099ebafd6a4a0fd1cf242d03c61b0253073e5 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/submit.sh @@ -0,0 +1,24 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +DRY_RUN="${DRY_RUN:-false}" +yaml_path="${SCRIPT_DIR}/train.yaml" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 +fi + +cmd=(snap_rutils cluster run -s "$yaml_path" -y) +echo "[submit] ${cmd[*]}" +if ! is_true "$DRY_RUN"; then + "${cmd[@]}" +fi diff --git a/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/train.yaml b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/train.yaml new file mode 100644 index 0000000000000000000000000000000000000000..42e1f3f54e9a3cd844d8869cc2f3c3ea4863ec16 --- /dev/null +++ b/deploy/yw006-4-sage-qwen35-9b-token-credit-video-adv-window5/train.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: True +priority: gpu-high + +pytorchjob: + replicas: 8 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && bash yw006-4-sage-qwen35-9b-token-credit-video-adv-window5.sh" + + jobs: + - name: ywang29-hunyuan-sage-qwen35-9b-tcva-w5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/submit_vbench2_generate_shards.sh b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_0.yaml b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fc8e8f2ea2c0fb3a26f0b15abbbe94e8dfc1b26f --- /dev/null +++ b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_1.yaml b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9efa6546e874a864fe55ea4c03a773b4e97614dc --- /dev/null +++ b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_2.yaml b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7c6590aefb7a9aaf785b9a3cc18b13f7a5165946 --- /dev/null +++ b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_3.yaml b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fef48251c8e5b891fb25a197ed70f4dcdcd50d05 --- /dev/null +++ b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_4.yaml b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bded50e68d4f462b7210f3f6e34763fabb3f7aea --- /dev/null +++ b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_5.yaml b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..15c426cfa94aeecd7199f55014ca8e2046332043 --- /dev/null +++ b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_6.yaml b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..12fc97c3f529e7f641d2b4bfb5c29297900cd5cf --- /dev/null +++ b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_7.yaml b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a3422bf43beca7110bfc976c7a34569b75c6ea72 --- /dev/null +++ b/deploy/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-7-sage-qwen35-9b-token-credit-per-frame-ratio 100" + + jobs: + - name: ywang29-vbench2-gen-z-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/job_name_random_suffix.txt b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..c1965706ae9a3bc9f2cc4de0bf80b57ec37cad66 --- /dev/null +++ b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/job_name_random_suffix.txt @@ -0,0 +1 @@ +rb6d4b6 diff --git a/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/submit_vbench2_generate_shards.sh b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_0.yaml b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5aa013244cd55c4665af29d79dd736ee11284b6e --- /dev/null +++ b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-gsva-rb6d4b6-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_1.yaml b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c73fc4038d0d9b703f7469596f4c57bd5f38342c --- /dev/null +++ b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-gsva-rb6d4b6-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_2.yaml b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..6ace85b9e6c2b65cdb401f7affe6c6318f023136 --- /dev/null +++ b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-gsva-rb6d4b6-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_3.yaml b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8456fd2cebf111b30a864e855afa54f27697e601 --- /dev/null +++ b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-gsva-rb6d4b6-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_4.yaml b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ca00b40b8f8104e3c37cfa2a4b0995abe8b452ca --- /dev/null +++ b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-gsva-rb6d4b6-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_5.yaml b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..768b18337d68647434d222711171cdaf771fa819 --- /dev/null +++ b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-gsva-rb6d4b6-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_6.yaml b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d0a3bfe5e400d789931567a68c0b033fa432666a --- /dev/null +++ b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-gsva-rb6d4b6-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_7.yaml b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fff59e9624222bfca741f58bcbd9e21051eee4ab --- /dev/null +++ b/deploy/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw006-8-sage-qwen35-9b-token-credit-general-semantic-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-gsva-rb6d4b6-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-1-sage-videoscore2-token-credit-video-adv-window5/submit.sh b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv-window5/submit.sh new file mode 100644 index 0000000000000000000000000000000000000000..dcf099ebafd6a4a0fd1cf242d03c61b0253073e5 --- /dev/null +++ b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv-window5/submit.sh @@ -0,0 +1,24 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +DRY_RUN="${DRY_RUN:-false}" +yaml_path="${SCRIPT_DIR}/train.yaml" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 +fi + +cmd=(snap_rutils cluster run -s "$yaml_path" -y) +echo "[submit] ${cmd[*]}" +if ! is_true "$DRY_RUN"; then + "${cmd[@]}" +fi diff --git a/deploy/yw008-1-sage-videoscore2-token-credit-video-adv-window5/train.yaml b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv-window5/train.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e99b8ed42dee68dbd16610792a2212fa0b5f1b2c --- /dev/null +++ b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv-window5/train.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: True +priority: gpu-high + +pytorchjob: + replicas: 8 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && bash yw008-1-sage-videoscore2-token-credit-video-adv-window5.sh" + + jobs: + - name: hunyuan-sage-vs2-tcva-abl-w5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/job_name_random_suffix.txt b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..8a7eed9e5eb498df74766f0a96d3b9572be4327b --- /dev/null +++ b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/job_name_random_suffix.txt @@ -0,0 +1 @@ +rd3c95b diff --git a/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/submit_vbench2_generate_shards.sh b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_0.yaml b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ed7d0f67fbbaa5a2320e125255f9d105c60eb0e2 --- /dev/null +++ b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs2-rd3c95b-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_1.yaml b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a12744034b97b1ad4a132482248ad7c3c2ae1c64 --- /dev/null +++ b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs2-rd3c95b-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_2.yaml b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..2654404145eeddb98f302f3a566948b3c2beef85 --- /dev/null +++ b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs2-rd3c95b-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_3.yaml b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0cab7b96f7cacbcf5445de636cdb22d23c502d85 --- /dev/null +++ b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs2-rd3c95b-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_4.yaml b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a7e12c7286ab17c90290d80279d45bb9aeff7bce --- /dev/null +++ b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs2-rd3c95b-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_5.yaml b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e15e06725906f91c43da6674873283e95b7d71b6 --- /dev/null +++ b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs2-rd3c95b-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_6.yaml b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..03b2e173aabf3f56205a5acbcf64db4e3452acbc --- /dev/null +++ b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs2-rd3c95b-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_7.yaml b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..412909d29851349fbf0a2019520547f165f048e2 --- /dev/null +++ b/deploy/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-1-sage-videoscore2-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-1-sage-videoscore2-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-vs2-rd3c95b-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/job_name_random_suffix.txt b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..6de86f3b8b3023896d0a3fcd82279cdcd22fe6b1 --- /dev/null +++ b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/job_name_random_suffix.txt @@ -0,0 +1 @@ +r73eb73 diff --git a/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/submit_vbench2_generate_shards.sh b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_0.yaml b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0745185392826d3f2eb88e3e61660814ace704e5 --- /dev/null +++ b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_1.yaml b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1be5faebf8090ba0047e2ecb812ee9d0e0af88c5 --- /dev/null +++ b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_2.yaml b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f482e714f05155786bb167d6bd08d1f2ae257b97 --- /dev/null +++ b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_3.yaml b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..89b1cf79b5972ade0c6b3ae14f07e1bd2f389bd8 --- /dev/null +++ b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_4.yaml b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..738326c2bbef780ca64fb05f2963f6edbe3db1b6 --- /dev/null +++ b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_5.yaml b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4a5db9c04b02d15560df3c1be866193f236e9a9e --- /dev/null +++ b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_6.yaml b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e8f2664552f3188339534981da826c2e6cf33da4 --- /dev/null +++ b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_7.yaml b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..27493f0bfdfd22f21503611a0849132e9e68ab4a --- /dev/null +++ b/deploy/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw008-3-sage-videoscore2-token-credit-video-adv-shuffle-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-vs23sw-r73eb73-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/job_name_random_suffix.txt b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..d601e72a04df37eee697692193084d1f688de997 --- /dev/null +++ b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/job_name_random_suffix.txt @@ -0,0 +1 @@ +r74e9a4 diff --git a/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/submit_vbench2_generate_shards.sh b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_0.yaml b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7d0aba9fbda9493f27668e5b0e1ac3018e9af7e4 --- /dev/null +++ b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw009-sage-internvl3-1b-token-credit-video-adv-2 100" + + jobs: + - name: ywang29-vbench2-gen-ivl3tcva2-r74e9a4-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_1.yaml b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..cb00e164c3ba1f42930923e6fa712c4ba0bb538c --- /dev/null +++ b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw009-sage-internvl3-1b-token-credit-video-adv-2 100" + + jobs: + - name: ywang29-vbench2-gen-ivl3tcva2-r74e9a4-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_2.yaml b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d2e76e84b55d96338cc6d436066c9e3947334acc --- /dev/null +++ b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw009-sage-internvl3-1b-token-credit-video-adv-2 100" + + jobs: + - name: ywang29-vbench2-gen-ivl3tcva2-r74e9a4-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_3.yaml b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7aa5828a367618371521aed55af87ac59f08b8d7 --- /dev/null +++ b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw009-sage-internvl3-1b-token-credit-video-adv-2 100" + + jobs: + - name: ywang29-vbench2-gen-ivl3tcva2-r74e9a4-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_4.yaml b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..19a66715f857608f32d98799372e79afb5e2848f --- /dev/null +++ b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw009-sage-internvl3-1b-token-credit-video-adv-2 100" + + jobs: + - name: ywang29-vbench2-gen-ivl3tcva2-r74e9a4-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_5.yaml b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1e5b7af02ad3a53eb498abf14845cc387dce7eac --- /dev/null +++ b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw009-sage-internvl3-1b-token-credit-video-adv-2 100" + + jobs: + - name: ywang29-vbench2-gen-ivl3tcva2-r74e9a4-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_6.yaml b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ee42d40728246aed2f1d1eb60649ab596d6b108c --- /dev/null +++ b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw009-sage-internvl3-1b-token-credit-video-adv-2 100" + + jobs: + - name: ywang29-vbench2-gen-ivl3tcva2-r74e9a4-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_7.yaml b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d76d58451b2dc2a315f22f75a5b30fb5647280ad --- /dev/null +++ b/deploy/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw009-sage-internvl3-1b-token-credit-video-adv-2/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw009-sage-internvl3-1b-token-credit-video-adv-2 100" + + jobs: + - name: ywang29-vbench2-gen-ivl3tcva2-r74e9a4-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/job_name_random_suffix.txt b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..5f5824f22542fb146444f616fe6cfaaf1ee8d182 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/job_name_random_suffix.txt @@ -0,0 +1 @@ +rd79c36 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/submit_vbench2_generate_shards.sh b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_0.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5ce3db0f1dca2e4ffe142f61540ae9818e8f5633 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-ur2sw-rd79c36-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_1.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8642f141a3454ffb8959ba6a1a92d87d98d1dc44 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-ur2sw-rd79c36-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_2.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..acc05735b7cd2c77e8fc306b1ff08ff11c428694 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-ur2sw-rd79c36-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_3.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5e4a259d24bc79ab0a0e72e42c205cd813d70e40 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-ur2sw-rd79c36-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_4.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f77e148738de077ce6b787a44b750e0d05283111 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-ur2sw-rd79c36-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_5.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c1aff5b9823f23254cf7f856a71a2725774c3a4b --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-ur2sw-rd79c36-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_6.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..80557bbe68595babba2274cda530d163ec99d199 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-ur2sw-rd79c36-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_7.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8e4af58ed4effde3b907f0775852bf187abe7d82 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv-shuffled-weighting 100" + + jobs: + - name: ywang29-vbench2-gen-ur2sw-rd79c36-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/job_name_random_suffix.txt b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/job_name_random_suffix.txt new file mode 100644 index 0000000000000000000000000000000000000000..bc4eccd8856eaf2fe35c009d533584073945e077 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/job_name_random_suffix.txt @@ -0,0 +1 @@ +rdf4d4e diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/submit_vbench2_generate_shards.sh b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/submit_vbench2_generate_shards.sh new file mode 100644 index 0000000000000000000000000000000000000000..bc4fdcd3b8ef69b680ed0e292f253eec1bbffe71 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/submit_vbench2_generate_shards.sh @@ -0,0 +1,29 @@ +#!/usr/bin/env bash +set -Eeuo pipefail + +SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +SHARD_COUNT="${SHARD_COUNT:-8}" +START_INDEX="${START_INDEX:-0}" +END_INDEX="${END_INDEX:-$((SHARD_COUNT - 1))}" +DRY_RUN="${DRY_RUN:-false}" + +is_true() { + case "${1:-}" in + true|TRUE|1|yes|YES|y|Y) return 0 ;; + *) return 1 ;; + esac +} + +for shard_index in $(seq "$START_INDEX" "$END_INDEX"); do + yaml_path="${SCRIPT_DIR}/vbench2_generate_shard_${shard_index}.yaml" + if [[ ! -f "$yaml_path" ]]; then + echo "[ERROR] Missing YAML: $yaml_path" >&2 + exit 1 + fi + + cmd=(snap_rutils cluster run -s "$yaml_path" -y) + echo "[submit] ${cmd[*]}" + if ! is_true "$DRY_RUN"; then + "${cmd[@]}" + fi +done diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_0.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_0.yaml new file mode 100644 index 0000000000000000000000000000000000000000..46c4d0e0b094d7ebcdcecb9560d73d80a2eaa75e --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_0.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=0 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-ur2tcva-rdf4d4e-shard-0 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_1.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..69f666e6e638eb008b1db334822b1e1ab0922655 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_1.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=1 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-ur2tcva-rdf4d4e-shard-1 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_2.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0600b1fdb66ecaddb7fd46f0f83b13b9903e8098 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_2.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=2 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-ur2tcva-rdf4d4e-shard-2 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_3.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bcc82425aa94782cb140374ef2f8f1da5c087e1a --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_3.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=3 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-ur2tcva-rdf4d4e-shard-3 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_4.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_4.yaml new file mode 100644 index 0000000000000000000000000000000000000000..3f1a154b7a6de018041548bf8966ef7233e51746 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_4.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=4 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-ur2tcva-rdf4d4e-shard-4 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_5.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_5.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a62b1b339ae6c2cef9ce1f674ae3b9012b36f987 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_5.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=5 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-ur2tcva-rdf4d4e-shard-5 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_6.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_6.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f32c351d45c18fc7106319ffcfae11875cef8eff --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_6.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=6 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-ur2tcva-rdf4d4e-shard-6 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2 diff --git a/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_7.yaml b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_7.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8f03612caba9f48e91b26ff9e9a782ea7ef89ff7 --- /dev/null +++ b/deploy/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_generate_shard_7.yaml @@ -0,0 +1,52 @@ +gpu_type: nvidia-tesla-a100 +gpu_num: "8" +gpu_memory: "80" +cpu_num: "88" +memory: "880.0" + +efa: False +priority: gpu-high + +pytorchjob: + replicas: 1 + +custom_node_labels: + snap.com/zone: us-west-2c + snap.com/ami_type: al2023 + +volcano_queue: mobile-video-backbone +script: + pre_run_event: "cd /nfs/ywang29/SnapVideo && /nfs/ywang29/Reward_finetuning/SAGE-GRPO/pre_run_2_10.sh && wandb login --relogin --host=https://api.wandb.ai 24d6afefd46915ea938dca1af6caf95ae0288a19" + command: "cd /nfs/ywang29/Reward_finetuning/SAGE-GRPO/ && EVAL_MODE=standard DIMENSIONS=Camera_Motion,Complex_Plot,Human_Anatomy,Human_Identity,Human_Clothes,Diversity,Composition,Dynamic_Spatial_Relationship,Dynamic_Attribute,Motion_Order_Understanding,Human_Interaction,Complex_Landscape,Motion_Rationality,Instance_Preservation,Mechanics,Thermotics,Material,Multi-View_Consistency MAX_PROMPTS_PER_DIM=0 SAMPLES_PER_PROMPT=auto PROMPT_SHARD_COUNT=8 PROMPT_SHARD_INDEX=7 PROMPT_SHARD_MODE=weighted BASE_RUN_DIR=/nfs/ywang29/Reward_finetuning/SAGE-GRPO/outputs/yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv/vbench2_eval/step_100 N_INFERENCE_GPU=auto bash run_sage_vbench2_generate.sh yw010-sage-unifiedreward2-qwen35-9b-token-credit-video-adv 100" + + jobs: + - name: ywang29-vbench2-gen-ur2tcva-rdf4d4e-shard-7 + +project_name: mobile_video +project_support_alias: ywang29 +team: creative_vision +cost_team: generative_ai +cost_feature: generative_ai_development +cost_sub_feature: video_generation + +docker_image: 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-training:2.9.0-gpu-py312-cu130-ubuntu22.04-ec2 + +mount: + - type: s3 + name: snap-research-cv-code + mount_path: /s3-code + - type: fsx + name: "fs-0aec9ce19ecc66a45" + mount_path: /nfs + - type: fsx + name: fs-0a730854e147a7bc2 + mount_path: /nfs-generative-one-shot + - type: fsx + name: fs-0edcd5941d62ce319 + mount_path: /snaphub + - type: s3 + name: snap-genvid + mount_path: /snap-genvid + - type: s3 + name: snap-genvid-us-east-2 + mount_path: /snap-genvid-us-east-2