12月1月2月3月4月5月6月7月8月9月10月11月
周一周三周五
一年内 265370 次贡献
更少更多
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 sync-whisper.cpp-25-05-20 分支比较 2 提交 »86650c6ec8 sync : whisper.cppec4e16dba4 examples : add --print-confidence option to cli (whisper/3150)
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 sync-whisper.cpp-25-07-12 分支d579069be3 sync : whispper.cpp
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 sync-whisper.cpp-25-07-02 分支ea50f7ed0d sync : whisper.cpp
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 sync-whisper.cpp-25-05-01 分支比较 3 提交 »3350042bf0 sync : whisper.cpp36edebc852 whisper : add check that target name exists (whisper/3103)afe4322c64 ggml : suppress Windows compiler warnings (whisper/3075)
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 sync-whisper-25-04-25 分支比较 10 提交 »52443f03db sync : whisper.cppada161a3ee cuda : fix unused variable compile warning (whisper/0)489716ba99 opencl : remove obsolete files (skip) (#1200)8d90ea0dd0 sync : llama.cpp939718f7c9 metal : add memory pool for temp allocs (llama/12850)
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 sync-llama.cpp-25-10-11 分支比较 10 提交 »22d8601978 sync : llama.cpp99bb022687 ggml : Fix FP16 ELU positive branch (llama/16519)e573287570 ggml: Correct SVE implementation in ggml_vec_dot_f16_unroll (llama/16518)1077761d99 CUDA: faster tile FA, add oob checks, more HSs (llama/16492)9a238bf7f8 sync : llama.cpp
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 sync-llama.cpp-26-07-17 分支比较 10 提交 »bdb9e134e9 sync : llama.cpp7d0001b605 ggml-blas: default hadamard mul_mat to cpu routine (llama/25710)c1c95265c0 vulkan: Support Q2_0 (llama/25430)c990045f00 sycl: fix row calculation when K_QUANTS_PER_ITERATION is 1 (llama/25690)a5aa8bbf53 opencl: add ABS op (llama/25115)
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 sync-llama.cpp-25-07-28 分支比较 10 提交 »8ca55dacb5 sync : llama.cpp62f0d682d9 vulkan : add fp16 support for the conv_2d kernel (llama/14872)fccda47ae0 vulkan: skip empty set_rows to avoid invalid API usage (llama/14860)0d380cd776 Docs: add instructions for adding backends (llama/14889)d743740d6d HIP: Enable Matrix cores for MMQ Kernels, Enable stream-K for CDNA 3 (llama/14624)
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 sync-llama.cpp-25-06-01 分支比较 10 提交 »d0f7473c1b sync : llama.cpp8b09bced08 threading: support for GGML_SCHED_PRIO_LOW, update thread info on Windows to avoid throttling (llama/12995)8720cc1601 CUDA: add a prop in ggml_cuda_device_infor for distinguish iGPU or dGPU in cuda (#13856) (llama/13895)ad9d6e65d0 CUDA: fix typo in FlashAttention code (llama/13926)c6689ee5e0 sched : avoid changing cur_copy when a graph is already allocated (llama/13922)
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 sync-llama.cpp-25-05-19 分支比较 10 提交 »c6202093c3 sync : llama.cpp6e098c4f58 CANN: Support MOE Model MUL_MAT_ID (llama/13042)304e28872e cmake: use the current build config for vulkan-shaders-gen (llama/13595)a2edecaf5d vulkan: move common FA code to flash_attn_base.comp (llama/13556)f43e37a798 vulkan: use scalar FA rather than coopmat2 when N==1 (llama/13554)