| 2024-01-13 |
Georgi Gerganov | sync : ggml |
commit | commitdiff | tree |
| 2024-01-13 |
Georgi Gerganov | examples : adapt to metal API |
commit | commitdiff | tree |
| 2024-01-13 |
Johannes Gäßler | ggml: cache sin/cos for RoPE (llama/4908) |
commit | commitdiff | tree |
| 2024-01-13 |
Georgi Gerganov | metal : remove old API (llama/4919) |
commit | commitdiff | tree |
| 2024-01-13 |
Georgi Gerganov | metal : disable log for loaded kernels (llama/4794) |
commit | commitdiff | tree |
| 2024-01-13 |
texmex76 | gguf : fix potential infinite for-loop (llama/4600) |
commit | commitdiff | tree |
| 2024-01-13 |
Georgi Gerganov | metal : refactor kernel loading code (llama/4794) |
commit | commitdiff | tree |
| 2024-01-13 |
Johannes Gäßler | CUDA: faster q8_0 -> f16 dequantization (llama/4895) |
commit | commitdiff | tree |
| 2024-01-13 |
RhinoDevel | talk-llama : add optional CLI arg to set the bot name... |
commit | commitdiff | tree |
| 2024-01-13 |
james wolf | examples : add python example for transcription (#1744) |
commit | commitdiff | tree |
| 2024-01-13 |
Georgi Gerganov | whisper : load the model into multiple buffers of max... |
commit | commitdiff | tree |
| 2024-01-12 |
Georgi Gerganov | talk-llama : sync llama.cpp |
commit | commitdiff | tree |
| 2024-01-12 |
Georgi Gerganov | sync : ggml |
commit | commitdiff | tree |
| 2024-01-12 |
slaren | backend_sched : fix assignments |
commit | commitdiff | tree |
| 2024-01-12 |
slaren | llama : ggml-backend integration (llama/4766) |
commit | commitdiff | tree |
| 2024-01-12 |
Johannes Gäßler | CUDA: fix softmax compile for old CUDA versions (llama... |
commit | commitdiff | tree |
| 2024-01-12 |
Kawrakow | Importance Matrix calculation (llama/4861) |
commit | commitdiff | tree |
| 2024-01-12 |
Sơn Phan Trung | models : make all scripts to be POSIX Compliant (#1725) |
commit | commitdiff | tree |
| 2024-01-12 |
Georgi Gerganov | ggml : fix 32-bit ARM compat for IQ2_XS (#1758) |
commit | commitdiff | tree |
| 2024-01-12 |
Boris Bliznioukov | go : add SetInitialPrompt method to bindings (#1753) |
commit | commitdiff | tree |
| 2024-01-12 |
George Hindle | server : add more parameters to server api (#1754) |
commit | commitdiff | tree |
| 2024-01-12 |
Georgi Gerganov | whisper : fix segment length with params.no_timestamps... |
commit | commitdiff | tree |
| 2024-01-12 |
George Hindle | params : don't compute timestamps when not printing... |
commit | commitdiff | tree |
| 2024-01-11 |
Georgi Gerganov | talk-llama : sync llama.cpp |
commit | commitdiff | tree |
| 2024-01-11 |
Georgi Gerganov | swift : remove local ggml.h reference |
commit | commitdiff | tree |
| 2024-01-11 |
Georgi Gerganov | swift : track ggml release branch |
commit | commitdiff | tree |
| 2024-01-11 |
Georgi Gerganov | sync : ggml |
commit | commitdiff | tree |
| 2024-01-11 |
Georgi Gerganov | sync : llama.cpp |
commit | commitdiff | tree |
| 2024-01-11 |
Kawrakow | ggml : SOTA 2-bit quants (add IQ2_XS) (llama/4856) |
commit | commitdiff | tree |
| 2024-01-11 |
Paul Tsochantaris | metal : put encoder debug group behind a define (llama... |
commit | commitdiff | tree |
| 2024-01-11 |
Georgi Gerganov | metal : improve dequantize precision to match CPU ... |
commit | commitdiff | tree |
| 2024-01-11 |
Georgi Gerganov | ggml : fix vld1q_s8_x4 32-bit compat (llama/4828) |
commit | commitdiff | tree |
| 2024-01-11 |
Johannes Gäßler | CUDA: faster softmax via shared memory + fp16 math... |
commit | commitdiff | tree |
| 2024-01-11 |
Georgi Gerganov | metal : fix deprecation warning (ggml/690) |
commit | commitdiff | tree |
| 2024-01-11 |
Timothy Cronin | ggml : remove ggml_cpy_inplace and ggml_cont_inplace... |
commit | commitdiff | tree |
| 2024-01-11 |
Jack Mousseau | metal : wrap each operation in debug group (ggml/690) |
commit | commitdiff | tree |
| 2024-01-11 |
leejet | ggml : change GGML_MAX_NAME at compile time (ggml/682) |
commit | commitdiff | tree |
| 2024-01-11 |
Halalaluyafail3 | Fix execlp call (ggml/689) |
commit | commitdiff | tree |
| 2024-01-11 |
Kawrakow | SOTA 2-bit quants (llama/4773) |
commit | commitdiff | tree |
| 2024-01-11 |
Johannes Gäßler | CUDA: fixed redundant value dequantization (llama/4809) |
commit | commitdiff | tree |
| 2024-01-11 |
Konstantin... | ggml : use __builtin_amdgcn_sudot4 in __dp4a for gfx11... |
commit | commitdiff | tree |
| 2024-01-11 |
Georgi Gerganov | ggml : do not sched_yield when calling BLAS (llama... |
commit | commitdiff | tree |
| 2024-01-11 |
Georgi Gerganov | ggml : include stdlib.h before intrin.h (llama/4736) |
commit | commitdiff | tree |
| 2024-01-10 |
Alexandru Mariuti | swift : checkout ggml commit instead of branch (#1750) |
commit | commitdiff | tree |
| 2024-01-10 |
RhinoDevel | talk-llama : add optional Piper TTS support (#1749) |
commit | commitdiff | tree |
| 2024-01-08 |
Emmanuel Schmidbauer | server : add request path option(#1741) |
commit | commitdiff | tree |
| 2024-01-08 |
Georgi Gerganov | main : add cli option to disable system prints (#1740) |
commit | commitdiff | tree |
| 2024-01-07 |
Georgi Gerganov | server : fix server temperature + add temperature_inc... |
commit | commitdiff | tree |
| 2024-01-06 |
Georgi Gerganov | talk-llama : sync latest llama.cpp |
commit | commitdiff | tree |
| 2024-01-05 |
Georgi Gerganov | release : v1.5.4 |
commit | commitdiff | tree |
| 2024-01-05 |
Erik Scholz | fix : cuda order of synchronization when setting a... |
commit | commitdiff | tree |
| 2024-01-05 |
Georgi Gerganov | metal : switch back to default.metallib (ggml/681) |
commit | commitdiff | tree |
| 2024-01-05 |
Georgi Gerganov | ggml : fix q2_k bpw in comments (ggml/680) |
commit | commitdiff | tree |
| 2024-01-04 |
Yajing Tang | coreml : fix ANE optimized encoder (#1716) |
commit | commitdiff | tree |
| 2024-01-04 |
Georgi Gerganov | whisper.swiftui : add .gitignore |
commit | commitdiff | tree |
| 2024-01-04 |
Georgi Gerganov | whispser : reset the "batched" timings (#1721) |
commit | commitdiff | tree |
| 2024-01-03 |
Georgi Gerganov | release : v1.5.3 |
commit | commitdiff | tree |
| 2024-01-03 |
Ashraful Islam | swift : update Package.swift to use ggml as package... |
commit | commitdiff | tree |
| 2024-01-03 |
Finn Voorhees | ggml : add error handling to graph_compute (#1714) |
commit | commitdiff | tree |
| 2024-01-03 |
Georgi Gerganov | cuda : simplify expression |
commit | commitdiff | tree |
| 2024-01-03 |
Georgi Gerganov | cuda : mark I16 and I32 ops as unsupported |
commit | commitdiff | tree |
| 2024-01-03 |
Georgi Gerganov | metal : add kernel_get_rows_i32 |
commit | commitdiff | tree |
| 2024-01-03 |
Georgi Gerganov | metal : optimize ggml_mul_mat_id (faster Mixtral PP... |
commit | commitdiff | tree |
| 2024-01-03 |
Georgi Gerganov | metal : enable shader debugging (cmake option) (llama... |
commit | commitdiff | tree |
| 2024-01-03 |
Georgi Gerganov | ggml : add ggml_vdotq_s32 alias (llama/4715) |
commit | commitdiff | tree |
| 2024-01-03 |
Johannes Gäßler | CUDA: fixed tensor cores not being used on RDNA3 (llama... |
commit | commitdiff | tree |
| 2024-01-03 |
automaticcat | ggml : add ggml_cpu_has_avx_vnni() (llama/4589) |
commit | commitdiff | tree |
| 2024-01-03 |
Johannes Gäßler | CUDA: fix tensor core logic for Pascal and HIP (llama... |
commit | commitdiff | tree |
| 2024-01-03 |
hydai | cuda: fix vmm oom issue on NVIDIA AGX Orin (llama/4687) |
commit | commitdiff | tree |
| 2024-01-03 |
Guillaume Wenzek | ggml : extend ggml_get_rows, ggml_repeat, ggml_concat... |
commit | commitdiff | tree |
| 2024-01-03 |
Georgi Gerganov | scripts : fix sync order + metal sed |
commit | commitdiff | tree |
| 2024-01-02 |
Andreu Huguet | examples : fix WASM Stack Overflow (#1713) |
commit | commitdiff | tree |
| 2023-12-30 |
bobqianic | docker : fix the publishing of the CUDA Docker image... |
commit | commitdiff | tree |
| 2023-12-29 |
Georgi Gerganov | scripts : do not sync commits from this repo |
commit | commitdiff | tree |
| 2023-12-29 |
Tamotsu Takahashi | ci : build with CLBlast + ggml-opencl use GGML_API... |
commit | commitdiff | tree |
| 2023-12-29 |
bobqianic | whisper : replace `tensor->n_dims` with `ggml_n_dims... |
commit | commitdiff | tree |
| 2023-12-29 |
Georgi Gerganov | sync : ggml (VMM, sync-ggml-am, dotprod ARM fixes,... |
commit | commitdiff | tree |
| 2023-12-29 |
Dimo | download : fix large q5 model name (#1695) |
commit | commitdiff | tree |
| 2023-12-23 |
bobqianic | whisper : Replace WHISPER_PRINT_DEBUG with WHISPER_LOG_... |
commit | commitdiff | tree |
| 2023-12-22 |
Georgi Gerganov | sync : ggml (ggml_scale, ggml_row_size, etc.) (#1677) |
commit | commitdiff | tree |
| 2023-12-22 |
Chaoqun | docker : Dockerize whisper.cpp (#1674) |
commit | commitdiff | tree |
| 2023-12-21 |
bobqianic | CI : Add coverage for talk-llama when WHISPER_CUBLAS... |
commit | commitdiff | tree |
| 2023-12-21 |
bobqianic | examples : Revert CMakeLists.txt for talk-llama (#1669) |
commit | commitdiff | tree |
| 2023-12-21 |
bobqianic | cmake : set default CUDA architectures (#1667) |
commit | commitdiff | tree |
| 2023-12-19 |
Alfredo Montesinos | bench.py : add different large models (#1655) |
commit | commitdiff | tree |
| 2023-12-14 |
Georgi Gerganov | wchess : update README.md |
commit | commitdiff | tree |
| 2023-12-14 |
Georgi Gerganov | release : v1.5.2 |
commit | commitdiff | tree |
| 2023-12-14 |
Georgi Gerganov | wchess : update readme |
commit | commitdiff | tree |
| 2023-12-14 |
fraxy-v | wchess : whisper assisted chess (#1595) |
commit | commitdiff | tree |
| 2023-12-13 |
Georgi Gerganov | sync : ggml (Metal fixes, new ops, tests) (#1633) |
commit | commitdiff | tree |
| 2023-12-12 |
Kreijstal | cmake : target windows 8 or above for prefetchVirtualMe... |
commit | commitdiff | tree |
| 2023-12-10 |
Kreijstal | cmake : Fix bug in httplib.h for mingw (#1615) |
commit | commitdiff | tree |
| 2023-12-08 |
Finn Voorhees | metal : fix `ggml_metal_log` vargs (#1606) |
commit | commitdiff | tree |
| 2023-12-08 |
Georgi Gerganov | whisper.objc : disable timestamps for real-time transcr... |
commit | commitdiff | tree |
| 2023-12-08 |
Georgi Gerganov | whisper : more debug messages + fix fallback logic |
commit | commitdiff | tree |
| 2023-12-08 |
Georgi Gerganov | metal : fix soft_max kernel src1 argument (#1602) |
commit | commitdiff | tree |
| 2023-12-07 |
Georgi Gerganov | sync : ggml (new ops, new backend, etc) (#1602) |
commit | commitdiff | tree |
| 2023-12-05 |
Oleg Sidorov | server : pass max-len argument to the server (#1574) |
commit | commitdiff | tree |
| 2023-12-05 |
Finn Voorhees | ios : Remove `#if arch(arm)` check for using Metal... |
commit | commitdiff | tree |
| 2023-12-03 |
Digipom | ggml : Fix 32-bit compiler warning (#1575) |
commit | commitdiff | tree |
| next |
overview / pkg / ggml / sources / whisper.cpp / shortlog