-
Notifications
You must be signed in to change notification settings - Fork 369
- #899 · SolitaryThinker opened
on Nov 24, 2025 2
Issues
is:issue state:open
is:issue state:open
Issue creation is restricted in this repository
Search results
[Bug] Slow VideoGenerator loading
installationInstallation and setup issuesInstallation and setup issuesscope: attentionAttention backends (VSA, STA, Flash, etc.)Attention backends (VSA, STA, Flash, etc.)scope: dataData preprocessing, datasetsData preprocessing, datasetsscope: docsDocumentationDocumentationscope: inferenceInference pipeline, serving, CLIInference pipeline, serving, CLIscope: modelModel architecture (DiTs, encoders, VAEs)Model architecture (DiTs, encoders, VAEs)Status: Open.#1491 In hao-ai-lab/FastVideo;[Bug] Failed to install via pip
installationInstallation and setup issuesInstallation and setup issuesscope: attentionAttention backends (VSA, STA, Flash, etc.)Attention backends (VSA, STA, Flash, etc.)Status: Open.#1490 In hao-ai-lab/FastVideo;[Feature] Standardize recipes and other inference stuff across model x hardware matrix
enhancementNew feature or requestNew feature or requestscope: attentionAttention backends (VSA, STA, Flash, etc.)Attention backends (VSA, STA, Flash, etc.)scope: inferenceInference pipeline, serving, CLIInference pipeline, serving, CLItype: ciCI/CD infrastructureCI/CD infrastructuretype: perfPerformance improvementPerformance improvementStatus: Open.[Feature] [ci] Normalize performance component timing extraction for stage subclasses
scope: attentionAttention backends (VSA, STA, Flash, etc.)Attention backends (VSA, STA, Flash, etc.)scope: inferenceInference pipeline, serving, CLIInference pipeline, serving, CLIscope: infraCI, tests, Docker, buildCI, tests, Docker, buildscope: modelModel architecture (DiTs, encoders, VAEs)Model architecture (DiTs, encoders, VAEs)type: ciCI/CD infrastructureCI/CD infrastructuretype: perfPerformance improvementPerformance improvementStatus: Open.#1475 In hao-ai-lab/FastVideo;[RFC]: Avoid GPU-to-CPU output materialization when save_video=False and return_frames=False
performancePerformance and memory issuesPerformance and memory issuesscope: attentionAttention backends (VSA, STA, Flash, etc.)Attention backends (VSA, STA, Flash, etc.)scope: inferenceInference pipeline, serving, CLIInference pipeline, serving, CLIStatus: Open.#1474 In hao-ai-lab/FastVideo;[Feature] I2V Dreamverse Support
scope: attentionAttention backends (VSA, STA, Flash, etc.)Attention backends (VSA, STA, Flash, etc.)scope: inferenceInference pipeline, serving, CLIInference pipeline, serving, CLIscope: modelModel architecture (DiTs, encoders, VAEs)Model architecture (DiTs, encoders, VAEs)Status: Open.#1452 In hao-ai-lab/FastVideo;Wan-VACE推理example
scope: docsDocumentationDocumentationscope: modelModel architecture (DiTs, encoders, VAEs)Model architecture (DiTs, encoders, VAEs)Status: Open.#1435 In hao-ai-lab/FastVideo;[tracker] Open add-model PRs
installationInstallation and setup issuesInstallation and setup issuesplatformPlatform-specific (Windows/macOS)Platform-specific (Windows/macOS)scope: attentionAttention backends (VSA, STA, Flash, etc.)Attention backends (VSA, STA, Flash, etc.)scope: distributedSP, FSDP, USP, multi-nodeSP, FSDP, USP, multi-nodescope: docsDocumentationDocumentationscope: inferenceInference pipeline, serving, CLIInference pipeline, serving, CLIscope: modelModel architecture (DiTs, encoders, VAEs)Model architecture (DiTs, encoders, VAEs)scope: trainingTraining pipeline, methods, configsTraining pipeline, methods, configsStatus: Open.#1407 In hao-ai-lab/FastVideo;[bug]: HunyuanVideo-1.5 generation crashes at text_encoding.py:221 — 'list' object has no attribute 'strip'
installationInstallation and setup issuesInstallation and setup issuesperformancePerformance and memory issuesPerformance and memory issuesscope: attentionAttention backends (VSA, STA, Flash, etc.)Attention backends (VSA, STA, Flash, etc.)scope: dataData preprocessing, datasetsData preprocessing, datasetsscope: distributedSP, FSDP, USP, multi-nodeSP, FSDP, USP, multi-nodescope: docsDocumentationDocumentationscope: inferenceInference pipeline, serving, CLIInference pipeline, serving, CLIscope: modelModel architecture (DiTs, encoders, VAEs)Model architecture (DiTs, encoders, VAEs)scope: trainingTraining pipeline, methods, configsTraining pipeline, methods, configsStatus: Open.#1387 In hao-ai-lab/FastVideo;RMSNorm.forward_native doesn't unwrap DTensor weight under FSDP CPU offload
bugSomething isn't workingSomething isn't workinghelp wantedExtra attention is neededExtra attention is neededinstallationInstallation and setup issuesInstallation and setup issuesscope: attentionAttention backends (VSA, STA, Flash, etc.)Attention backends (VSA, STA, Flash, etc.)scope: distributedSP, FSDP, USP, multi-nodeSP, FSDP, USP, multi-nodescope: modelModel architecture (DiTs, encoders, VAEs)Model architecture (DiTs, encoders, VAEs)scope: trainingTraining pipeline, methods, configsTraining pipeline, methods, configsStatus: Open.#1379 In hao-ai-lab/FastVideo;[RFC]: Robust performance regression tracking enhancement
enhancementNew feature or requestNew feature or requestinstallationInstallation and setup issuesInstallation and setup issuesperformancePerformance and memory issuesPerformance and memory issuesplatformPlatform-specific (Windows/macOS)Platform-specific (Windows/macOS)scope: attentionAttention backends (VSA, STA, Flash, etc.)Attention backends (VSA, STA, Flash, etc.)scope: dataData preprocessing, datasetsData preprocessing, datasetsscope: distributedSP, FSDP, USP, multi-nodeSP, FSDP, USP, multi-nodescope: docsDocumentationDocumentationscope: inferenceInference pipeline, serving, CLIInference pipeline, serving, CLIscope: infraCI, tests, Docker, buildCI, tests, Docker, buildscope: kernelCUDA kernels, fastvideo-kernelCUDA kernels, fastvideo-kernelscope: modelModel architecture (DiTs, encoders, VAEs)Model architecture (DiTs, encoders, VAEs)scope: trainingTraining pipeline, methods, configsTraining pipeline, methods, configstype: ciCI/CD infrastructureCI/CD infrastructureStatus: Open.[Bug] Is the MoE expert routing in _sample_timesteps intended to be flow-shift-aware?
installationInstallation and setup issuesInstallation and setup issuesscope: attentionAttention backends (VSA, STA, Flash, etc.)Attention backends (VSA, STA, Flash, etc.)scope: distributedSP, FSDP, USP, multi-nodeSP, FSDP, USP, multi-nodescope: inferenceInference pipeline, serving, CLIInference pipeline, serving, CLIscope: modelModel architecture (DiTs, encoders, VAEs)Model architecture (DiTs, encoders, VAEs)scope: trainingTraining pipeline, methods, configsTraining pipeline, methods, configsStatus: Open.#1370 In hao-ai-lab/FastVideo;