cosmopolitan/third_party/ggml
Justine Tunney 791f79fcb3
Make improvements
- We now serialize the file descriptor table when spawning / executing
  processes on Windows. This means you can now inherit more stuff than
  just standard i/o. It's needed by bash, which duplicates the console
  to file descriptor #255. We also now do a better job serializing the
  environment variables, so you're less likely to encounter E2BIG when
  using your bash shell. We also no longer coerce environ to uppercase

- execve() on Windows now remotely controls its parent process to make
  them spawn a replacement for itself. Then it'll be able to terminate
  immediately once the spawn succeeds, without having to linger around
  for the lifetime as a shell process for proxying the exit code. When
  process worker thread running in the parent sees the child die, it's
  given a handle to the new child, to replace it in the process table.

- execve() and posix_spawn() on Windows will now provide CreateProcess
  an explicit handle list. This allows us to remove handle locks which
  enables better fork/spawn concurrency, with seriously correct thread
  safety. Other codebases like Go use the same technique. On the other
  hand fork() still favors the conventional WIN32 inheritence approach
  which can be a little bit messy, but is *controlled* by guaranteeing
  perfectly clean slates at both the spawning and execution boundaries

- sigset_t is now 64 bits. Having it be 128 bits was a mistake because
  there's no reason to use that and it's only supported by FreeBSD. By
  using the system word size, signal mask manipulation on Windows goes
  very fast. Furthermore @asyncsignalsafe funcs have been rewritten on
  Windows to take advantage of signal masking, now that it's much more
  pleasant to use.

- All the overlapped i/o code on Windows has been rewritten for pretty
  good signal and cancelation safety. We're now able to ensure overlap
  data structures are cleaned up so long as you don't longjmp() out of
  out of a signal handler that interrupted an i/o operation. Latencies
  are also improved thanks to the removal of lots of "busy wait" code.
  Waits should be optimal for everything except poll(), which shall be
  the last and final demon we slay in the win32 i/o horror show.

- getrusage() on Windows is now able to report RUSAGE_CHILDREN as well
  as RUSAGE_SELF, thanks to aggregation in the process manager thread.
2023-10-08 08:59:53 -07:00
..
common.cc Fix small matters and improve sysconf() 2023-08-17 00:32:11 -07:00
common.h Fix small matters and improve sysconf() 2023-08-17 00:32:11 -07:00
companionai.txt Upgrade llama.cpp to e6a46b0ed1884c77267dc70693183e3b7164e0e0 2023-05-10 04:20:48 -07:00
fp16.c Prevent ftrace from misaligning functions 2023-06-06 06:00:31 -07:00
fp16.h Make more ML improvements 2023-05-16 08:07:23 -07:00
fp16.internal.h Make more ML improvements 2023-05-16 08:07:23 -07:00
ggjt.v1.c Get radpajama to build 2023-05-13 20:44:36 -07:00
ggjt.v1.internal.h Add support for new GGJT v2 quantizers 2023-05-13 08:08:32 -07:00
ggjt.v1.q4_0.c Upgrade to Cosmopolitan GCC 11.2.0 for aarch64 2023-06-05 02:07:28 -07:00
ggjt.v1.q4_0.h Add support for new GGJT v2 quantizers 2023-05-13 08:08:32 -07:00
ggjt.v1.q4_1.c Upgrade to Cosmopolitan GCC 11.2.0 for aarch64 2023-06-05 02:07:28 -07:00
ggjt.v1.q4_1.h Add support for new GGJT v2 quantizers 2023-05-13 08:08:32 -07:00
ggjt.v1.q4_2.c Upgrade to Cosmopolitan GCC 11.2.0 for aarch64 2023-06-05 02:07:28 -07:00
ggjt.v1.q4_2.h Add support for new GGJT v2 quantizers 2023-05-13 08:08:32 -07:00
ggjt.v1.q5_0.c Upgrade to Cosmopolitan GCC 11.2.0 for aarch64 2023-06-05 02:07:28 -07:00
ggjt.v1.q5_0.h Add support for new GGJT v2 quantizers 2023-05-13 08:08:32 -07:00
ggjt.v1.q5_1.c Upgrade to Cosmopolitan GCC 11.2.0 for aarch64 2023-06-05 02:07:28 -07:00
ggjt.v1.q5_1.h Add support for new GGJT v2 quantizers 2023-05-13 08:08:32 -07:00
ggjt.v1.q8_0.c Upgrade to Cosmopolitan GCC 11.2.0 for aarch64 2023-06-05 02:07:28 -07:00
ggjt.v1.q8_0.h Add support for new GGJT v2 quantizers 2023-05-13 08:08:32 -07:00
ggjt.v1.q8_1.c Upgrade to Cosmopolitan GCC 11.2.0 for aarch64 2023-06-05 02:07:28 -07:00
ggjt.v1.q8_1.h Add support for new GGJT v2 quantizers 2023-05-13 08:08:32 -07:00
ggjt.v2.c Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggjt.v2.internal.h Upgrade to Cosmopolitan GCC 11.2.0 for aarch64 2023-06-05 02:07:28 -07:00
ggjt.v2.q4_0.c Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggjt.v2.q4_0.h Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggjt.v2.q4_1.c Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggjt.v2.q4_1.h Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggjt.v2.q5_0.c Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggjt.v2.q5_0.h Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggjt.v2.q5_1.c Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggjt.v2.q5_1.h Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggjt.v2.q8_0.c Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggjt.v2.q8_0.h Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggjt.v2.q8_1.c Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggjt.v2.q8_1.h Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggml.c Make improvements 2023-09-18 21:04:47 -07:00
ggml.h Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
ggml.mk Make improvements 2023-10-08 08:59:53 -07:00
LICENSE Import llama.cpp 2023-04-27 14:37:14 -07:00
llama.cc Fix warnings 2023-09-01 20:50:18 -07:00
llama.h Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00
llama_util.h Get us closer to building busybox 2023-06-18 04:13:45 -07:00
main.cc Fix warnings 2023-09-01 20:50:18 -07:00
perplexity.cc Perform some code cleanup 2023-05-15 16:32:10 -07:00
quantize.cc Fix small matters and improve sysconf() 2023-08-17 00:32:11 -07:00
README.cosmo Introduce support for GGJT v3 file format 2023-06-03 15:46:21 -07:00

DESCRIPTION

  ggml is a machine learning library useful for LLM inference on CPUs

LICENSE

  MIT

ORIGIN

  https://github.com/ggerganov/llama.cpp
  d8bd0013e8768aaa3dc9cfc1ff01499419d5348e

LOCAL CHANGES

  - Maintaining support for deprecated file formats
  - Make it possible for loaded prompts to be cached to disk
  - Introduce -v and --verbose flags
  - Reduce batch size from 512 to 32
  - Allow --n_keep to specify a substring of prompt
  - Don't print stats / diagnostics unless -v is passed
  - Reduce --top_p default from 0.95 to 0.70
  - Change --reverse-prompt to no longer imply --interactive
  - Permit --reverse-prompt specifying custom EOS if non-interactive
  - Refactor headers per cosmo convention
  - Remove C++ exceptions; use Die() function instead
  - Removed division from matrix multiplication.
  - Let quantizer convert between ggmt formats