Skip to content

Pull requests: nod-ai/shark-ai

New pull request New
52 Open 771 Closed
52 Open 771 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Move the vae reshaping logic back to its correct location
#986 opened Feb 19, 2025 by KyleHerndon Loading…
Enable llama fp8 masked_flash_attention 8
#984 opened Feb 19, 2025 by AmosLewis • Draft
Test CI
#981 opened Feb 19, 2025 by stellaraccident • Draft
Speed-up all_gather by reducing buffer copies.
#979 opened Feb 18, 2025 by Alex-Vasile Loading…
Improve T5 encoder tests with more prompts and static context length
#976 opened Feb 17, 2025 by sogartar • Draft
1
Bump IREE requirement pins to their latest versions.
#973 opened Feb 17, 2025 by shark-pr-automator bot Loading…
@renxida
3
[shortfin-sd] Add exports and support for scheduled unet, batch sizes.
#972 opened Feb 16, 2025 by monorimet • Draft
5
[sharktank][ci] Re-enable t5 data-dependent tests
#969 opened Feb 14, 2025 by renxida • Draft
@renxida
Don't use GGUF but convert T5 directly from Hugging Face
#967 opened Feb 14, 2025 by sogartar Loading…
2
Add metrics to sdxl phases (#948)
#961 opened Feb 12, 2025 by daveliddell Loading…
1
Reduce buffer copying by using one device to reduce and distribute
#957 opened Feb 11, 2025 by Alex-Vasile Loading…
1
Update nightly_releases.md
#949 opened Feb 11, 2025 by pdhirajkumarprasad Loading…
1
Update random generation for fp8 quantization
#945 opened Feb 10, 2025 by rsuderman Loading…
Adding conv igemm layout support - step 1: conv op matcher
#943 opened Feb 10, 2025 by jerryyin Loading…
5
Add sharding support for latent attention block
#935 opened Feb 7, 2025 by rsuderman Loading…
[CI][sharktank] Move Sharktank Data-Dependent Tests to OSSCI Cluster
#932 opened Feb 7, 2025 by renxida • Draft
3
Update llama benchmarking tests to llama instruct variants
#909 opened Feb 4, 2025 by aviator19941 Loading…
[DNM] cherry pick fp8 attn nonsense with hack cream
#907 opened Feb 4, 2025 by dan-garvey • Draft
2
Shard input_mask for Llama
#905 opened Feb 3, 2025 by stbaione • Draft
4
Adding conv igemm layout support
#898 opened Feb 3, 2025 by jerryyin Loading…
39
[sharktank] restore custom matmul kernel
#896 opened Feb 2, 2025 by dan-garvey Loading…
1
26
Make get_iree_devices read IREE_DEVICE env var if provided
#891 opened Jan 31, 2025 by sogartar Loading…
5
Make Flux Transformer RoPE use a custom IREE kernel
#871 opened Jan 27, 2025 by sogartar Loading…
7
Add Flux transformer benchmarking
#870 opened Jan 27, 2025 by sogartar Loading…
43
[Llama] Kv cache new layout
#858 opened Jan 22, 2025 by Groverkss • Draft
ProTip! no:milestone will show everything without a milestone.