What's Changed
- Add Dockerfile for ARM64 architecture support and update README instructions by @z4y4ts in #827
- feat: multi-arch CUDA Dockerfile and sm_121 (DGX Spark GB10) by @nazq in #840
- feat: update
metricspackage to support modern rust by @daeho-ro in #850 - Support
harrier-oss-v1model by @kozistr in #854 - Fix to only build
Dockerfile-cudaw/ Blackwell 12.1 on linux/arm64 by @alvarobartt in #852 - Add minimal AMD ROCm GPU support by @Abdennacer-Badaoui in #853
- 🔒 Pin GitHub Actions to commit SHAs by @paulinebm in #855
- Add dedicated linux/arm64 runner by @alvarobartt in #858
- [AMD] Add Instinct GPU setup guide by @Abdennacer-Badaoui in #856
- chore: bump doc-builder SHA for PR upload workflow by @rtrompier in #862
- fix(entrypoint): add /usr/local/nvidia to PATH and ldconfig on GKE by @oOraph in #877
- Add default user-agent header by @alvarobartt in #892
- Fix
max_position_embeddingshandling inNomicBertConfigandFlashNomicBertConfigby @alvarobartt in #876 - fix(docker): detect CUDA version from
CUDA UMD Versionheader on driver 6xx (#870) by @Anai-Guo in #871 - Align
cuda-{all-entrypoint,entrypoint}.shby @alvarobartt in #893 - DebertaV2 fix for running with large batches by @vrdn-23 in #846
- Fix link to AMD Instinct GPU guide in README by @manickavela-uni in #898
- Fix Cloud Run CLI argument documentation by @cupkk in #901
- Fix single predict request metrics by @original4422 in #917
- Quality: Unbounded buffer allocation in PaddedBatch.from_pb by @kumburovicbranko682-boop in #890
- feat: ROCm flash-attn varlen, triton layer norm, and AMD Dockerfile by @Abdennacer-Badaoui in #860
- Update
versionto 1.9.4 & runcargo updateby @alvarobartt in #928
New Contributors
- @z4y4ts made their first contribution in #827
- @nazq made their first contribution in #840
- @daeho-ro made their first contribution in #850
- @Abdennacer-Badaoui made their first contribution in #853
- @paulinebm made their first contribution in #855
- @rtrompier made their first contribution in #862
- @oOraph made their first contribution in #877
- @Anai-Guo made their first contribution in #871
- @manickavela-uni made their first contribution in #898
- @cupkk made their first contribution in #901
- @original4422 made their first contribution in #917
- @kumburovicbranko682-boop made their first contribution in #890
Full Changelog: v1.9.3...v1.9.4