Skip to content

Actions: NVIDIA/cosmos-framework

Actions

Pre-commit

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
pre-commit.yml
535 workflow runs
535 workflow runs

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
fix(inference): materialise checkpoint tensors in host memory before H2D copy (17.7x faster load on GB300) Pre-commit #573: Pull request #150 synchronize by lfengad
1m 12s rwagwani:fix/checkpoint-load-mmap-h2d
rwagwani:fix/checkpoint-load-mmap-h2d
1m 12s
Fix gradient accumulation under FSDP2: defer cross-replica all-reduce Pre-commit #572: Pull request #162 synchronize by lfengad
Action required Sullivan07043:fix-fsdp2-accum
Sullivan07043:fix-fsdp2-accum
Action required
Upgrade pinned hf CLI to 1.26.0 for Xet-backed repos Pre-commit #571: Pull request #163 synchronize by lfengad
1m 14s Sullivan07043:fix-hf-cli-xet
Sullivan07043:fix-hf-cli-xet
1m 14s
Update uv.lock (#165) Pre-commit #570: Commit 5e02e64 pushed by foreverlms
1m 4s main
main
1m 4s
Add TAO-compatible Cosmos3 fine-tuning support Pre-commit #569: Pull request #164 synchronize by ramanathan831
Action required ramanathan831:dev/ram/cosmos3-tao-reproducibility
ramanathan831:dev/ram/cosmos3-tao-reproducibility
Action required
Add TAO-compatible Cosmos3 fine-tuning support Pre-commit #568: Pull request #164 synchronize by ramanathan831
Action required ramanathan831:dev/ram/cosmos3-tao-reproducibility
ramanathan831:dev/ram/cosmos3-tao-reproducibility
Action required
Update uv.lock Pre-commit #567: Pull request #165 synchronize by foreverlms
1m 2s fix_docker
fix_docker
1m 2s
Update uv.lock Pre-commit #566: Pull request #165 synchronize by foreverlms
2m 7s fix_docker
fix_docker
2m 7s
Update uv.lock Pre-commit #565: Pull request #165 opened by foreverlms
1m 36s fix_docker
fix_docker
1m 36s
Add TAO-compatible Cosmos3 fine-tuning support Pre-commit #564: Pull request #164 synchronize by ramanathan831
Action required ramanathan831:dev/ram/cosmos3-tao-reproducibility
ramanathan831:dev/ram/cosmos3-tao-reproducibility
Action required
feat: load ModelOpt static-FP8 checkpoints via TorchAO Pre-commit #563: Pull request #157 synchronize by pengcuo
1m 14s feat/modelopt-fp8-checkpoint-loading
feat/modelopt-fp8-checkpoint-loading
1m 14s
Add TAO-compatible Cosmos3 fine-tuning support Pre-commit #562: Pull request #164 synchronize by ramanathan831
Action required ramanathan831:dev/ram/cosmos3-tao-reproducibility
ramanathan831:dev/ram/cosmos3-tao-reproducibility
Action required
Add TAO-compatible Cosmos3 fine-tuning support Pre-commit #561: Pull request #164 opened by ramanathan831
Action required ramanathan831:dev/ram/cosmos3-tao-reproducibility
ramanathan831:dev/ram/cosmos3-tao-reproducibility
Action required
fix(inference): materialise checkpoint tensors in host memory before H2D copy (17.7x faster load on GB300) Pre-commit #560: Pull request #150 synchronize by rwagwani
1m 2s rwagwani:fix/checkpoint-load-mmap-h2d
rwagwani:fix/checkpoint-load-mmap-h2d
1m 2s
Upgrade pinned hf CLI to 1.26.0 for Xet-backed repos Pre-commit #559: Pull request #163 opened by Sullivan07043
1m 8s Sullivan07043:fix-hf-cli-xet
Sullivan07043:fix-hf-cli-xet
1m 8s
Fix gradient accumulation under FSDP2: defer cross-replica all-reduce Pre-commit #558: Pull request #162 opened by Sullivan07043
1m 22s Sullivan07043:fix-fsdp2-accum
Sullivan07043:fix-fsdp2-accum
1m 22s
fix(inference): materialise checkpoint tensors in host memory before H2D copy (17.7x faster load on GB300) Pre-commit #557: Pull request #150 synchronize by rwagwani
1m 13s rwagwani:fix/checkpoint-load-mmap-h2d
rwagwani:fix/checkpoint-load-mmap-h2d
1m 13s
feat(vlm): LoRA SFT support for VLM + VideoPhy-2 LoRA recipes Pre-commit #556: Pull request #155 synchronize by lfengad
1m 7s liangf/vlm-lora-sft
liangf/vlm-lora-sft
1m 7s
feat(vlm): LoRA SFT support for VLM + VideoPhy-2 LoRA recipes Pre-commit #555: Pull request #155 synchronize by lfengad
1m 15s liangf/vlm-lora-sft
liangf/vlm-lora-sft
1m 15s
feat(vlm): LoRA SFT support for VLM + VideoPhy-2 LoRA recipes Pre-commit #554: Pull request #155 synchronize by lfengad
1m 1s liangf/vlm-lora-sft
liangf/vlm-lora-sft
1m 1s
inference/action: let forward dynamics run for per-dataset action wid… Pre-commit #553: Commit ee58e41 pushed by lfengad
1m 0s main
main
1m 0s
feat(vlm): LoRA SFT support for VLM + VideoPhy-2 LoRA recipes Pre-commit #552: Pull request #155 synchronize by lfengad
1m 7s liangf/vlm-lora-sft
liangf/vlm-lora-sft
1m 7s
feat(vlm): LoRA SFT support for VLM + VideoPhy-2 LoRA recipes Pre-commit #551: Pull request #155 synchronize by lfengad
1m 4s liangf/vlm-lora-sft
liangf/vlm-lora-sft
1m 4s
fix(inference): materialise checkpoint tensors in host memory before H2D copy (17.7x faster load on GB300) Pre-commit #550: Pull request #150 synchronize by rwagwani
-1s rwagwani:fix/checkpoint-load-mmap-h2d
rwagwani:fix/checkpoint-load-mmap-h2d
-1s
inference/action: let forward dynamics run for per-dataset action widths Pre-commit #549: Pull request #159 opened by lfengad
1m 8s liangf/action-fd-per-dataset-action-width
liangf/action-fd-per-dataset-action-width
1m 8s