Title: [Dev] Implement TileLang NF Format Dequantize by LeiWang1999 · Pull Request #273 · microsoft/BitBLAS · GitHub
Open Graph Title: [Dev] Implement TileLang NF Format Dequantize by LeiWang1999 · Pull Request #273 · microsoft/BitBLAS
X Title: [Dev] Implement TileLang NF Format Dequantize by LeiWang1999 · Pull Request #273 · microsoft/BitBLAS
Description: BitBLAS is a library to support mixed-precision matrix multiplications, especially for quantized LLM deployment. - [Dev] Implement TileLang NF Format Dequantize by LeiWang1999 · Pull Request #273 · microsoft/BitBLAS
Open Graph Description: I think we should still enable the NF format to achieve optimal performance when utilizing fp16xnf4 matrix multiplication, especially with atomicAdd enabled.”
X Description: I think we should still enable the NF format to achieve optimal performance when utilizing fp16xnf4 matrix multiplication, especially with atomicAdd enabled.”
Opengraph URL: https://github.com/microsoft/BitBLAS/pull/273
X: @github
Domain: github.com
| route-pattern | /_view_fragments/voltron/pull_requests/show/:user_id/:repository/:id/pull_request_layout(.:format) |
| route-controller | voltron_pull_requests_fragments |
| route-action | pull_request_layout |
| fetch-nonce | v2:d92befc7-c032-1411-826f-aaf00b455041 |
| current-catalog-service-hash | ae870bc5e265a340912cde392f23dad3671a0a881730ffdadd82f2f57d81641b |
| request-id | E7DC:3C0BF2:46770:5DF28:6A4C11E3 |
| html-safe-nonce | c6c0d3df344cd443d355f032b502a06b8caf58dad89830e1d410efa72ab3d7d4 |
| visitor-payload | eyJyZWZlcnJlciI6IiIsInJlcXVlc3RfaWQiOiJFN0RDOjNDMEJGMjo0Njc3MDo1REYyODo2QTRDMTFFMyIsInZpc2l0b3JfaWQiOiIyODU2MTUwMDE5NTEyOTk2MzIzIiwicmVnaW9uX2VkZ2UiOiJpYWQiLCJyZWdpb25fcmVuZGVyIjoiaWFkIn0= |
| visitor-hmac | e3e53526ee411f542107c9ecbfdb52b0265727c92dbde3f0f7dd447ffabd6d0a |
| hovercard-subject-tag | pull_request:2241883667 |
| github-keyboard-shortcuts | repository,pull-request-list,pull-request-conversation,pull-request-files-changed,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/voltron/pull_requests/show/microsoft/BitBLAS/273/pull_request_layout |
| twitter:image | https://opengraph.githubassets.com/5ecb600c2237f29fdf426b231e44b4e7f11c7b898d7425d704ae21981857d7e5/microsoft/BitBLAS/pull/273 |
| twitter:card | summary_large_image |
| og:image | https://opengraph.githubassets.com/5ecb600c2237f29fdf426b231e44b4e7f11c7b898d7425d704ae21981857d7e5/microsoft/BitBLAS/pull/273 |
| og:image:alt | I think we should still enable the NF format to achieve optimal performance when utilizing fp16xnf4 matrix multiplication, especially with atomicAdd enabled.” |
| og:image:width | 1200 |
| og:image:height | 600 |
| og:site_name | GitHub |
| og:type | object |
| og:author:username | LeiWang1999 |
| hostname | github.com |
| expected-hostname | github.com |
| None | 0ccfc9e5281bfe12e38a4d632dc422843e4d5b6757917f7efda2f6567d72fea9 |
| turbo-cache-control | no-cache |
| go-import | github.com/microsoft/BitBLAS git https://github.com/microsoft/BitBLAS.git |
| octolytics-dimension-user_id | 6154722 |
| octolytics-dimension-user_login | microsoft |
| octolytics-dimension-repository_id | 762181983 |
| octolytics-dimension-repository_nwo | microsoft/BitBLAS |
| octolytics-dimension-repository_public | true |
| octolytics-dimension-repository_is_fork | false |
| octolytics-dimension-repository_network_root_id | 762181983 |
| octolytics-dimension-repository_network_root_nwo | microsoft/BitBLAS |
| 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 | 18812876666a23a0912556e224383baa4c84cf8e |
| ui-target | full |
| theme-color | #1e2327 |
| color-scheme | light dark |
Links:
Viewport: width=device-width