use fprintf for diagnostic output, keep printf only for printing model output#48
use fprintf for diagnostic output, keep printf only for printing model output#48ggerganov merged 1 commit intoggml-org:masterfrom prusnak:info-stderr
Conversation
ggerganov
left a comment
ggerganov
left a comment
There was a problem hiding this comment.
No need to modify ggml.c and ggml.h - it does not print anything.
Keep only the rest of the changes
keep printf only for printing model output one can now use ./main ... 2>dev/null to suppress any diagnostic output
|
Rebased on top of current master, kept only changes in main.cpp |
|
|
Hello |
|
What does this mean? The command 2>dev/null is used to redirect the standard error (stderr) output of a command to /dev/null, which is a special file in Unix-like systems that discards any data written to it. This is often used to suppress error messages or diagnostic output that may be displayed on the terminal. In your case, it seems like there is a typo in the command. The correct syntax is /dev/null with a slash before dev, not dev/null. So the corrected command would be: javascript This should redirect the stderr output of the ./chat command to /dev/null, effectively suppressing any diagnostic output that may be displayed on the terminal. In Windows PowerShell, you can achieve a similar result of redirecting stderr output to null using the > operator to redirect the output stream and 2>&1 to redirect the error stream to the output stream. Here's an example command: java In this example, 2>&1 redirects the stderr stream to the stdout stream and > $null redirects the stdout stream to $null, which is a PowerShell automatic variable representing a null or empty value. This effectively discards any output from the command, including both stdout and stderr streams. Note that the path separators in Windows PowerShell use a backslash () instead of a forward slash (/), and the executable extension for Windows is .exe instead of no extension. |
Yes, there was a typo in my original post, it should read /dev/null, I updated the post. |
|
Thanks, this helps when I was writing a function to call alpaca.cpp/llama.cpp from JS |
MSVC Compatibility fix for timer
ggml-org#48) * This commit changes the naming conventioins of the shared object files. This modificatioin will produce libfalcon.so instead of liblibfalcon.so which would create downstreem bugs. * This commit renames the falcon.cpp and falcon.h files the libfalcon.cpp and libfalcon.h. It produces shared object files that maintain the standard file format of libfalcon.so
…gml-org#48) To trigger the pipeline, add the audio-release label to the PR. PR must target tarek/feat/audio2
* First version of AVX2 Flash attention I simply took the Zen4 implementation and converted platform specific stuff to methods of a struct providing data loading/storing, conversions, multiply, add, etc. Most likely not optimal as the Zen4 strategy has been designed based on having 32 512-bit registers, so basically we can have 4X more data stored in vector registers compared to AVX2 with 16 x 256-bit. It still gives a small speedup (~4% at 2048 tokens) for Gemma-2b. * Fix Zenn4 parts broken via the AVX2 change * Try smaller q_step - no improvement * Fix ARM_NEON I had forgotten to guard the AVX2/Zen4 implementation against __aarch64__ --------- Co-authored-by: Iwan Kawrakow <[email protected]>
4 participants
one can use
./main ... 2>/dev/nullto suppress any diagnostic outputFixes #5