aboutsummaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2023-07-14ggml : fix static_assert with older compilers #2024 (#2218)Evan Miller
2023-07-14llama : add functions that work directly on model (#2197)Bach Le
* Remove vocab reference from context * Add functions that works directly with model
2023-07-14build.zig : install config header (#2216)Ali Chraghi
2023-07-14examples : fixed path typos in embd-input (#2214)Shangning Xu
2023-07-14cuda : support broadcast add & mul (#2192)Jiahao Li
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-07-14CUDA: mul_mat_vec_q kernels for k-quants (#2203)Johannes Gäßler
2023-07-14make : fix combination of LLAMA_METAL and LLAMA_MPI (#2208)James Reynolds
Fixes https://github.com/ggerganov/llama.cpp/issues/2166 by moving commands after the CFLAGS are changed.
2023-07-14ggml : sync (ggml_conv_2d, fix mul_mat bug, CUDA GLM rope)Georgi Gerganov
2023-07-14Metal: faster Q4_0 and Q4_1 matrix x vector kernels (#2212)Kawrakow
* 3-5% faster Q4_0 on Metal * 7-25% faster Q4_1 on Metal * Oops, forgot to delete the original Q4_1 kernel --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-07-13Revert "Support using mmap when applying LoRA (#2095)" (#2206)Howard Su
Has perf regression when mlock is used. This reverts commit 2347463201a9f4159ae95b737e1544dd300569c8.
2023-07-13Fix compile error on Windows CUDA (#2207)Howard Su
2023-07-13devops : add missing quotes to bash script (#2193)Bodo Graumann
This prevents accidentally expanding arguments that contain spaces.
2023-07-12metal : new q4_0 matrix-vector kernel (#2188)Shouzheng Liu
Prefetch data to improve GPU utilization. ~48% faster for 33B model.
2023-07-12ggml : broadcast mul_mat + conv batch support (#2199)Georgi Gerganov
* ggml : broadcast mul_mat + conv batch support * ggml : apply mul_mat broadcast fix by @jploski
2023-07-12ggml : add ggml_pool_1d and ggml_pool_2dGeorgi Gerganov
2023-07-12cuda : add gelu supportGeorgi Gerganov
2023-07-12FP16 is supported in CM=6.0 (#2177)Howard Su
* FP16 is supported in CM=6.0 * Building PTX code for both of 60 and 61 Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
2023-07-12Fixed __dp4a compute capability: 6.0 -> 6.1 (#2189)Johannes Gäßler
2023-07-12ggml : revert CUDA broadcast changes from #2183 (#2191)Georgi Gerganov
2023-07-11ggml : sync (abort callback, mul / add broadcast, fix alibi) (#2183)Georgi Gerganov
2023-07-11ggml : remove src0 and src1 from ggml_tensor and rename opt to src (#2178)Spencer Sutton
* Add ggml changes * Update train-text-from-scratch for change * mpi : adapt to new ggml_tensor->src --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-07-11llama : add classifier-free guidance (#2135)Bach Le
* Initial implementation * Remove debug print * Restore signature of llama_init_from_gpt_params * Free guidance context * Make freeing of guidance_ctx conditional * Make Classifier-Free Guidance a sampling function * Correct typo. CFG already means context-free grammar. * Record sampling time in llama_sample_classifier_free_guidance * Shift all values by the max value before applying logsoftmax * Fix styling based on review
2023-07-11docker : add '--server' option (#2174)Jinwoo Jeong
2023-07-11readme : fix zig build instructions (#2171)Chad Brewbaker
2023-07-11Support using mmap when applying LoRA (#2095)Howard Su
* Support using mmap when applying LoRA * Fix Linux * Update comment to reflect the support lora with mmap
2023-07-11Possible solution to allow K-quants on models with n_vocab!=32000 (#2148)LostRuins
* This allows LLAMA models that were previously incompatible with K quants to function mostly as normal. This happens when a model has a vocab != 32000, e.g 32001 which means it's not divisible by 256 or 64. Since the problematic dimensions only apply for `tok_embeddings.weight` and `output.weight` (dimentions 4096 x n_vocab), we can simply quantize these layers to Q8_0 whereas the majority of the hidden layers are still K-quanted since they have compatible dimensions. * Fix indentation Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * As an alternative, to avoid failing on Metal due to lack of Q8_0 support, instead quantize tok_embeddings.weight to Q4_0 and retain output.weight as F16. This results in a net gain of about 55mb for a 7B model compared to previous approach, but should minimize adverse impact to model quality. --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-07-10mpi : add support for distributed inference via MPI (#2099)Evan Miller
* MPI support, first cut * fix warnings, update README * fixes * wrap includes * PR comments * Update CMakeLists.txt * Add GH workflow, fix test * Add info to README * mpi : trying to move more MPI stuff into ggml-mpi (WIP) (#2099) * mpi : add names for layer inputs + prep ggml_mpi_graph_compute() * mpi : move all MPI logic into ggml-mpi Not tested yet * mpi : various fixes - communication now works but results are wrong * mpi : fix output tensor after MPI compute (still not working) * mpi : fix inference * mpi : minor * Add OpenMPI to GH action * [mpi] continue-on-error: true * mpi : fix after master merge * [mpi] Link MPI C++ libraries to fix OpenMPI * tests : fix new llama_backend API * [mpi] use MPI_INT32_T * mpi : factor out recv / send in functions and reuse * mpi : extend API to allow usage with outer backends (e.g. Metal) --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-07-09llama : remove "first token must be BOS" restriction (#2153)oobabooga
2023-07-09main : escape prompt prefix/suffix (#2151)Nigel Bosch
2023-07-09readme : update Termux instructions (#2147)JackJollimore
The file pathing is significant when running models inside of Termux on Android devices. llama.cpp performance is improved with loading a .bin from the $HOME directory.
2023-07-09ggml : fix buidling with Intel MKL but ask for "cblas.h" issue (#2104) (#2115)clyang
* Fix buidling with Intel MKL but ask for "cblas.h" issue * Use angle brackets to indicate the system library
2023-07-09readme : add more docs indexes (#2127)rankaiyx
* Update README.md to add more docs indexes * Update README.md to add more docs indexes
2023-07-08Fixed OpenLLaMA 3b CUDA mul_mat_vec_q (#2144)Johannes Gäßler
2023-07-08CUDA: add __restrict__ to mul mat vec kernels (#2140)Johannes Gäßler
2023-07-07docker : add support for CUDA in docker (#1461)dylan
Co-authored-by: canardleteer <eris.has.a.dad+github@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-07-07ci : switch threads to 1 (#2138)Georgi Gerganov
2023-07-07ggml : change ggml_graph_compute() API to not require context (#1999)Qingyou Meng
* ggml_graph_compute: deprecate using ggml_context, try resolve issue #287 * rewrite: no longer consider backward compitability; plan and make_plan * minor: rename ctx as plan; const * remove ggml_graph_compute from tests/test-grad0.c, but current change breaks backward * add static ggml_graph_compute_sugar() * minor: update comments * reusable buffers * ggml : more consistent naming + metal fixes * ggml : fix docs * tests : disable grad / opt + minor naming changes * ggml : add ggml_graph_compute_with_ctx() - backwards compatible API - deduplicates a lot of copy-paste * ci : enable test-grad0 * examples : factor out plan allocation into a helper function * llama : factor out plan stuff into a helper function * ci : fix env * llama : fix duplicate symbols + refactor example benchmark * ggml : remove obsolete assert + refactor n_tasks section * ggml : fix indentation in switch * llama : avoid unnecessary bool * ggml : remove comments from source file and match order in header --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-07-07ggml : remove sched_yield() call in ggml_graph_compute_thread() (#2134)Georgi Gerganov
2023-07-07convert.py: add mapping for safetensors bf16 (#1598)Aarni Koskela
Fixes #1473
2023-07-07Fix opencl by wrap #if-else-endif with \n (#2086)Howard Su
2023-07-06ggml : fix restrict usageGeorgi Gerganov
2023-07-06convert : update for baichuan (#2081)Judd
1. guess n_layers; 2. relax warnings on context size; 3. add a note that its derivations are also supported. Co-authored-by: Judd <foldl@boxvest.com>
2023-07-06alpaca.sh : update model file name (#2074)tslmy
The original file name, `ggml-alpaca-7b-q4.bin`, implied the first-generation GGML. After the breaking changes (mentioned in https://github.com/ggerganov/llama.cpp/issues/382), `llama.cpp` requires GGML V3 now. Those model files are named `*ggmlv3*.bin`. We should change the example to an actually working model file, so that this thing is more likely to run out-of-the-box for more people, and less people would waste time downloading the old Alpaca model.
2023-07-05Expose generation timings from server & update completions.js (#2116)Tobias Lütke
* use javascript generators as much cleaner API Also add ways to access completion as promise and EventSource * export llama_timings as struct and expose them in server * update readme, update baked includes * llama : uniform variable names + struct init --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-07-05Update Server Instructions (#2113)Jesse Jojo Johnson
* Update server instructions for web front end * Update server README * Remove duplicate OAI instructions * Fix duplicate text --------- Co-authored-by: Jesse Johnson <thatguy@jessejojojohnson.com>
2023-07-05ggml : fix bug introduced in #1237Georgi Gerganov
2023-07-05tests : fix test-grad0Georgi Gerganov
2023-07-05ggml : generalize `quantize_fns` for simpler FP16 handling (#1237)Stephan Walter
* Generalize quantize_fns for simpler FP16 handling * Remove call to ggml_cuda_mul_mat_get_wsize * ci : disable FMA for mac os actions --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-07-05Update server instructions for web front end (#2103)Jesse Jojo Johnson
Co-authored-by: Jesse Johnson <thatguy@jessejojojohnson.com>
2023-07-05Quantized dot products for CUDA mul mat vec (#2067)Johannes Gäßler
generated by cgit v1.2.3 (git 2.25.1) at 2024-10-18 21:20:25 +0000