为了安全，强烈建议开启2FA双因子认证：User Settings -> Account -> Enable two-factor authentication！！！

Tags

Tags give the ability to mark specific points in history as being important

b1434

fdee152e · starcoder : add GPU offloading (#3827) · Oct 28, 2023
b1433

41aee4df · speculative : ensure draft and target model vocab matches (#3812) · Oct 28, 2023
b1432

6d459cbf · llama : correctly report GGUFv3 format (#3818) · Oct 27, 2023
b1431

c8d6a1f3 · simple : fix batch handling (#3803) · Oct 27, 2023
b1430

2f9ec7e2 · cuda : improve text-generation and batched decoding performance (#3776) · Oct 27, 2023
b1429

34b2a5e1 · server : do not release slot on image input (#3798) · Oct 26, 2023
b1428

6961c4bd · batched-bench : print params at start · Oct 25, 2023
b1427

cc448774 · log : disable pid in log filenames · Oct 25, 2023
b1426

ad939626 · server : add parameter -tb N, --threads-batch N (#3584) (#3768) · Oct 24, 2023
b1425

1717521c · server : do not block system prompt update (#3767) · Oct 24, 2023
b1424

b2f7e04b · sync : ggml (conv ops + cuda MSVC fixes) (#3765) · Oct 24, 2023
b1423

abd21fc9 · cmake : add missed dependencies (#3763) · Oct 24, 2023
b1422

2b4ea35e · cuda : add batched cuBLAS GEMM for faster attention (#3749) · Oct 24, 2023
b1421

daab3d7f · Add more tokenizer tests (#3742) · Oct 24, 2023
b1420

469c9add · metal : handle ggml_scale for n%4 != 0 (close #3754) · Oct 24, 2023
b1419

e3932593 · Revert "make : add optional CUDA_NATIVE_ARCH (#2482)" · Oct 23, 2023
b1416

5be6c803 · llama : remove token functions with `context` args in favor of `model` (#3720) · Oct 23, 2023
b1414

96981f37 · make : add optional CUDA_NATIVE_ARCH (#2482) · Oct 22, 2023
b1413

438c2ca8 · server : parallel decoding and multimodal (#3677) · Oct 22, 2023
b1412

9e70cc03 · Add test for MPT tokenization (#3728) · Oct 22, 2023