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-llama.cpp-25-05-13 分支比较 10 提交 »7b00719840 examples : update to ggml-opt and ggml-backend changes (#0)9333635a6e sync : llama.cppb00cc55911 opencl: remove unnecessary assert for `add` (llama/13257)153827fac0 llama/ggml: add LLM training support (llama/10544)4fbd64950d ggml-cpu: Integrate fp32=bf16xbf16 SME KleidiAI kernel (llama/13053)
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 sync-llama.cpp-25-03-27-2 分支比较 10 提交 »93c462c240 sync : llama.cppab9cc55a4c scripts : update sync (#1161)660def0639 files : remove old wkv6 sources (#0)c407e1476c sync : llama.cpp2e727b4be9 ggml : sync/merge cmake,riscv,powerpc, add common.cmake (#0)
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 sync-llama.cpp-25-03-31 分支比较 10 提交 »ea78ccd608 sync : llama.cppab67db5cff SYCL: Remove misleading ggml_sycl_op_flatten function (llama/12387)81247ee8d6 metal : use constexpr in FA kernels + fix typedef (llama/12659)028d33c66a musa: fix all warnings, re-enable `-DLLAMA_FATAL_WARNINGS=ON` in ci and update doc (llama/12611)38709cf089 cmake : fix ccache conflict (llama/12522)
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 sync-llama.cpp-24-12-12-0 分支比较 10 提交 »b374c321e8 sync : llama.cpp8fb4f536b8 ggml: load all backends from a user-provided search path (llama/10699)e0d06b747a vulkan: request round-to-even for fp16 in im2col/rope_head (llama/10767)2653d31ce3 vulkan: dynamic subgroup size for the remaining k quants (llama/10745)9320c279f8 CUDA: rename macros to avoid conflicts with WinAPI (llama/10736)
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 gq 分支比较 10 提交 »724c45d545 ggml : finalize the Q4_1 quantization for ARM_NEON154fcc3a2b ggml : sync all changes from llama.cpp and whisper.cpp331a8621c7 utils : print quantization histogramsc4f14039f4 ggml : add WASM SIMD for Q4_0a85bc0fd61 whisper : add Q4_1 model sizes
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 llama 分支比较 10 提交 »7f32376b70 llama : initial working FP16 + 4-bit Q4_0c2e9635c79 llama : initial model conversion to ggml format3adf02e311 utils : print quantization histograms05e7d26ba4 ggml : add WASM SIMD for Q4_03f08ce7004 whisper : add Q4_1 model sizes
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 mul-mat-id-batch 分支比较 10 提交 »e715793edf only run GGML_TASK_INIT and GGML_TASK_FINALIZE once4bde984585 add metal impl8b6f8a9261 ggml : add support multiple ids rows in ggml_mat_mul_id95cdaf9d7c sync : whisper.cpp (metal soft max fix + example prints)c57aa8e905 sync : llama.cpp (fused soft max, gpu cpy ops, etc.) (#640)
- frostbyte_neo 推送了仓库 github-featured/ggml-org--ggml 的 release 分支比较 10 提交 »03c1ad89ca sync : whisper.cpp3faad1efed examples : added audio_ctx argument to main and server (whisper/1857)a05a28fa2b metal : option to embed MSL source into compiled binary (whisper/1842)055b4c2989 examples : initialize context params properly (whisper/1852)15438356ac sync : llama.cpp