Commit History
-
Update README.md
Verified signature6748a1f
-
attn to cross_attn key conversion (#27)
- attn to cross_attn key conversion (aec6c7f79e76042b50ec724810bd877e3497bd18)4748682 -
Update model_index.json
Verified signature7c840a1 -
Update model_index.json
Verified signatured83bd85 -
Update README.md
Verified signaturef1508f8 -
Create guider/guider_config.json
Verified signatured15715f
-
Update README.md
Verified signature5d8cdc7
-
attn key conversion (#25)
- Update attn keys (e7fc5136d70afa8d2d68f03ccfd9f49a70e2dad0)861efe8 -
dpmsolver (#24)
- Update README.md (de4d0dd57a1657e36ea929bca42a9520800dcc47)6ae296e -
Update README.md
Verified signature9152c1e -
diffusers integration (#23)
- Update diffusers integration (83dc22d9db931b0d98183ee86be4414690192817) - Delete inference.py (15d68f9137246214d1b2afc84e52cc2d90bf2636) - Update README.md (d2cb24cdd7482c46ac0c3842d73c29b9e227c042)7acae4b -
docs: add RTX 4090 benchmark + GPU arch list for SageAttention build (#22)
- docs: add RTX 4090 benchmark + GPU arch list for SageAttention build (2302f6bd85ac3271b69db57d0ddb9853448a18b4)637b205
-
docs: split Memory-efficient Inference and GGUF+SageAttention into sub READMEs (#20)
- docs: split Memory-efficient Inference and GGUF+SageAttention into sub READMEs (e9a061b3abf9db713413ddef437adf948fd2364c) - docs: minor README formatting adjustment (9a9a0f47e6835678d0930df94814234ee60af452)190632e -
docs: ComfyUI release + section update (#18)
- docs: add ComfyUI release to News + update ComfyUI section (b790e27480a8df0a0366aac876accfc88dcaa77e)191ad8c -
fix: readme anchor links for HF renderer
Verified signatureb59003f -
fix: readme anchor links for HF renderer
Verified signatureb187480 -
fix: readme anchor links for HF renderer
Verified signature55e6b6a -
feat: GGUF + SageAttention guide — DPMSolver++ default, benchmark, README update (#17)
- feat: GGUF + SageAttention guide — DPMSolver++ default, benchmark, README update (922a8b78f59f5ad9d0d114f4390d5a1b224e1f06) - Upload assets/sage_compare_BF16.webp with huggingface_hub (b1f39ad47ba4b4fc30f0936e9d2375b0e6e48594) - Upload assets/sage_compare_Q4_K_M.webp with huggingface_hub (81665a89a57553481cbe2db7573e5cab87a1a212) - Upload assets/sage_compare_Q5_K_M.webp with huggingface_hub (c188e4526e859d2f477688719ced04ca804b0f8c) - Upload assets/sage_compare_Q8_0.webp with huggingface_hub (0ca01ed010c3983fb8809d4ad0f81030ff2ef4b6)ed9bedb
-
Delete assets/vows.mp4
Verified signature87d296f -
Update README.md
Verified signaturee736003 -
Upload woman.mp4
Verified signature8ba0cb3 -
Update README.md
Verified signature7f55c2f -
Upload 5 files
Verified signatureb24c8d1
-
Upload i2v_sample.jpg
Verified signature3b63fd7
-
Add FP8 weight quantization guide to README (#9)
- docs: add FP8 weight quantization guide to Memory-efficient Inference (2e572b381ffd25b29760e0f51a0c5618caaae5d5)dd6b2de
-
Enable Flash Attention by trimming prompt embedding padding (#8)
- feat: trim prompt_embeds padding for batch=1 to enable Flash Attention (d1a46b18645d57467e630bf664f2e9f34cb92d6d) - feat: move prompt_embeds trim to __call__ for correct CFG alignment (eabf2fae153ad8ecc6d9837c55fefdc8d025bddc) - feat: skip encoder_attention_mask in guider when mask is None (2134e9d09511f37876026af8d08e9f8ea20fa850) - chore: verify batch>1 preserves original attention_mask path (item 4) (8ca961d2d0ce0acbb589012dc8bee4d266b0b1bd) - chore: verify I2V compatibility with encoder_attention_mask=None (item 5) (9717caf550404f4d91c8cadf65e0adcb3c018161) - docs: add Flash Attention section to README (4b01b7406caf46637b3c541e1beb47f58cd4c2a2) - fix: trim pos/neg independently instead of max alignment (b67e1aacb2ad744282f9fcc364ce8706070087c0) - chore: remove debug prints and Flash Attention README section (20f9f0572b691df5ffde509bd72b4d9c83741380)3a95e87 -
Delete motif-video-technical-report.pdf
Verified signature0e8fceb -
Update README.md
Verified signature68210f0
-
Update README.md
Verified signature82fb28f -
Switch to T5Gemma2Encoder for text encoding (#5)
- Switch to T5Gemma2Encoder for text encoding (c625c21f6ed24a2ea7c2895ce671acec0de66f31)cb8fbd9 -
chore: remove .plans from tracking (internal working docs)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>c863b63
-
chore: close low-vram-inference plan (MM-1091)
All checklist items completed. HF pushed, Jira closed. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>0ed7f12 -
feat: enable_model_cpu_offload() for 24GB GPUs + README update
- Fix T5Gemma2Model device mismatch in pipeline_motif_video.py (accelerate hook on parent, but .encoder child called directly) - Add Memory-efficient Inference section to README (expandable_segments:True + peak ~19GB on nvidia-smi) - Update Requirements: 30GB+ for default, 24GB with offload - Remove Scheduler shift from Recommended Settings (auto-configured) - Add ComfyUI coming soon section Measured on H200: allocated 14.66GB, nvidia-smi peak ~19GB with expandable_segments:True (vs ~25GB without). Jira: MM-1091 Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>160bda0
-
Update README.md
Verified signature8114af8 -
add-tech-report (#2)
- Update techreport link to Motifvideo_techreport.pdf (c92fbc618b8e557b36aef6b0aff1118f7fa3810c) - Update tech report link to motif-video-technical-report.pdf (d68b84329e0d7115c148267f0576514493e35b92) - Merge main into pr/2 and resolve links conflicts (a7b750b62bbef6be098d92d2b9e43a3ebacff89b)3e1bd47 -
Update techreport link to Motifvideo_techreport.pdf
beomgyub8007d8 -
Update README.md
Verified signaturea92ae5e -
Upload motif-video-technical-report.pdf
Verified signature97d9c21 -
Update README.md
Verified signature6b6fae5
-
Tech report coming soon
c629900 -
Upload assets/i2v_sample.jpg with huggingface_hub
Verified signature80745e9 -
Remove leftover comment, fix copyright year
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>2caebc1 -
Upload assets/i2v_sample.jpg with huggingface_hub
Verified signature4acd806 -
Initial release
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>54cdb04
-
initial commit
Verified signature874fd5a