Title: Consider using pre-built Flash Attention kernels via `kernels` · Issue #319 · OpenGVLab/InternVideo · GitHub
Open Graph Title: Consider using pre-built Flash Attention kernels via `kernels` · Issue #319 · OpenGVLab/InternVideo
X Title: Consider using pre-built Flash Attention kernels via `kernels` · Issue #319 · OpenGVLab/InternVideo
Description: Hey, I am Sayak from the Kernels team at Hugging Face. I noticed that this project uses Flash Attention which includes a long build time. We ship pre-built binaries (which provide bit-exact outputs as the upstream) and thereby, we make i...
Open Graph Description: Hey, I am Sayak from the Kernels team at Hugging Face. I noticed that this project uses Flash Attention which includes a long build time. We ship pre-built binaries (which provide bit-exact outputs...
X Description: Hey, I am Sayak from the Kernels team at Hugging Face. I noticed that this project uses Flash Attention which includes a long build time. We ship pre-built binaries (which provide bit-exact outputs...
Opengraph URL: https://github.com/OpenGVLab/InternVideo/issues/319
X: @github
Domain: github.com
{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Consider using pre-built Flash Attention kernels via `kernels`","articleBody":"Hey,\n\nI am Sayak from the [Kernels](https://github.com/huggingface/kernels) team at Hugging Face. I noticed that this project uses Flash Attention which includes a long build time. We ship pre-built binaries (which provide bit-exact outputs as the upstream) and thereby, we make it easy to use.\n\nUsing FA3 on a supported machine is as easy as:\n\n```py\n# make sure `kernels` is installed: `pip install -U kernels`\nfrom kernels import get_kernel\n\nkernel_module = get_kernel(\"kernels-community/flash-attn3\")\nflash_attn_func = kernel_module.flash_attn_func\n\nflash_attn_func(...)\n```\n\nLet us know if you'd be interested in this and and we'd be happy to provide a draft of how it would look in your repo.\n","author":{"url":"https://github.com/sayakpaul","@type":"Person","name":"sayakpaul"},"datePublished":"2026-05-28T03:17:20.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":0},"url":"https://github.com/319/InternVideo/issues/319"}
| route-pattern | /_view_fragments/issues/show/:user_id/:repository/:id/issue_layout(.:format) |
| route-controller | voltron_issues_fragments |
| route-action | issue_layout |
| fetch-nonce | v2:9d0d42bc-4ff6-286c-7b45-a5eaaf2c4568 |
| current-catalog-service-hash | 81bb79d38c15960b92d99bca9288a9108c7a47b18f2423d0f6438c5b7bcd2114 |
| request-id | 9504:3EA9A9:2FE5D76:4453B04:6A4C0C7A |
| html-safe-nonce | 11cf563560f668b645210b98469fe3332a8291daa93dd73306a4e78ec3b356ec |
| visitor-payload | eyJyZWZlcnJlciI6IiIsInJlcXVlc3RfaWQiOiI5NTA0OjNFQTlBOToyRkU1RDc2OjQ0NTNCMDQ6NkE0QzBDN0EiLCJ2aXNpdG9yX2lkIjoiNzE4ODIwNjAyODg1NjIzMzA4MiIsInJlZ2lvbl9lZGdlIjoiaWFkIiwicmVnaW9uX3JlbmRlciI6ImlhZCJ9 |
| visitor-hmac | c40fea9d861e13de87a4607f421ec841d296dcdcfd627f7d57603adca1dde3f9 |
| hovercard-subject-tag | issue:4537458939 |
| github-keyboard-shortcuts | repository,issues,copilot |
| google-site-verification | Apib7-x98H0j5cPqHWwSMm6dNU4GmODRoqxLiDzdx9I |
| octolytics-url | https://collector.github.com/github/collect |
| analytics-location | / |
| fb:app_id | 1401488693436528 |
| apple-itunes-app | app-id=1477376905, app-argument=https://github.com/_view_fragments/issues/show/OpenGVLab/InternVideo/319/issue_layout |
| twitter:image | https://opengraph.githubassets.com/0c25f570670b47c6c8398f32156296a3031c263100f2f2ba34c9a43660a8e000/OpenGVLab/InternVideo/issues/319 |
| twitter:card | summary_large_image |
| og:image | https://opengraph.githubassets.com/0c25f570670b47c6c8398f32156296a3031c263100f2f2ba34c9a43660a8e000/OpenGVLab/InternVideo/issues/319 |
| og:image:alt | Hey, I am Sayak from the Kernels team at Hugging Face. I noticed that this project uses Flash Attention which includes a long build time. We ship pre-built binaries (which provide bit-exact outputs... |
| og:image:width | 1200 |
| og:image:height | 600 |
| og:site_name | GitHub |
| og:type | object |
| og:author:username | sayakpaul |
| hostname | github.com |
| expected-hostname | github.com |
| None | 940b0dbb4d0fe72558ee760b2ff0f25a504383fdfe404f3a7691ec853b8dc17f |
| turbo-cache-control | no-preview |
| go-import | github.com/OpenGVLab/InternVideo git https://github.com/OpenGVLab/InternVideo.git |
| octolytics-dimension-user_id | 94522163 |
| octolytics-dimension-user_login | OpenGVLab |
| octolytics-dimension-repository_id | 569713120 |
| octolytics-dimension-repository_nwo | OpenGVLab/InternVideo |
| octolytics-dimension-repository_public | true |
| octolytics-dimension-repository_is_fork | false |
| octolytics-dimension-repository_network_root_id | 569713120 |
| octolytics-dimension-repository_network_root_nwo | OpenGVLab/InternVideo |
| turbo-body-classes | logged-out env-production page-responsive |
| disable-turbo | false |
| browser-stats-url | https://api.github.com/_private/browser/stats |
| browser-errors-url | https://api.github.com/_private/browser/errors |
| release | b0267d17ea43e55446153a7e2c60bb6ac3cffd6b |
| ui-target | full |
| theme-color | #1e2327 |
| color-scheme | light dark |
Links:
Viewport: width=device-width