Tags

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

b2346

61d1c88e · Vulkan Improvements (#5835) · Mar 05, 2024
b2345

21b08674 · [SYCL] fix mul_mat fault in CI/unit-test (#5862) · Mar 05, 2024
b2343

29eee404 · fix speculative decoding build on windows (#5874) · Mar 04, 2024
b2334

4ffcdce2 · add alias for chat template (#5858) · Mar 04, 2024
b2333

a0fc6266 · sync : ggml · Mar 04, 2024
b2331

82f3e668 · common : use LLAMA_DEFAULT_SEED (#5855) · Mar 04, 2024
b2330

5a51cc1b · main : support special tokens as reverse/anti prompt (#5847) · Mar 04, 2024
b2329

67be2ce1 · cuda : fix data race in soft max (#5853) · Mar 03, 2024
b2327

475df1d6 · llama : allow for user specified embedding pooling type (#5849) · Mar 03, 2024
b2325

de9692a7 · llama : fix llama_copy_state_data with fragmented KV cache (#5840) · Mar 03, 2024
b2324

e6029348 · ci : schedule slow server tests only on Release or on demand (#5839) · Mar 03, 2024
b2323

8ef969af · server : init http requests thread pool with --parallel if set (#5836) · Mar 03, 2024
b2321

97311342 · server: tests: passkey challenge / self-extend with context shift demo (#5832) · Mar 02, 2024
b2320

4a6e2d61 · llama : add abort_callback to interrupt computation (#5409) · Mar 02, 2024
b2319

494c8703 · ggml : fix IQ3_S AVX implementation (#5834) · Mar 02, 2024
b2318

4d4d2366 · convert : automatically fall back to HfVocab if tokenizer.model doesn't exist (#5821) · Mar 02, 2024
b2316

bbde6eb2 · ggml : IQ3_S improvements (#5829) · Mar 02, 2024
b2314

6c32d8c7 · llama : refactor internal quantization functions (#5830) · Mar 02, 2024
b2313

802da009 · llama : fix segfault from unknown model arch name (#5820) · Mar 02, 2024
b2312

71564139 · Support multiple GPUs (split mode) on SYCL backend (#5806) · Mar 02, 2024

Prev
1
…
19
20
21
22
23
24
25
26
27
…
99
Next