v0.0.1-beta61

What's Changed

• Flash attention support for Llama by @dvruette in #2277
• Use fixed RNG seed value for all DeepSpeed workers by @andreaskoepf in #2324
• Show model config in the chat UI by @AbdBarho in #2317
• Populate env vars before render layout by @notmd in #2323
• Update configs according to feedback to #2277 by @dvruette in #2325
• Disable Initial Prompt Task for en and es Locales by @hzj5790 in #1849
• stats are shown on the admin by @vivasvan1 in #2330
• inference: use uuid v7 for most of table by @notmd in #2327
• fix deepspeed issue on trainer_rm.py, add crossentropy support by @theblackcat102 in #2321
• Improve scores of small 1.4B reward model.. by @andreaskoepf in #2329
• Refactor inference backend auth and switch to authlib by @olliestanley in #2318
• add dataset counts script by @CloseChoice in #2294
• fix typos by @RainRat in #2264
• Updated nginx config for prod, including streaming headers by @yk in #2239

New Contributors

• @vivasvan1 made their first contribution in #2330
• @RainRat made their first contribution in #2264

Full Changelog: v0.0.1-beta60...v0.0.1-beta61