libLBFGS 1.4 release candidate.
git-svn-id: file:///home/svnrepos/software/liblbfgs/trunk@6 ecf4c44f-38d1-4fa4-9757-a0b4dd0349fc
This commit is contained in:
parent
09554fbf50
commit
423f64c6cc
2
COPYING
2
COPYING
@ -1,7 +1,7 @@
|
|||||||
The MIT License
|
The MIT License
|
||||||
|
|
||||||
Copyright (c) 1990 Jorge Nocedal
|
Copyright (c) 1990 Jorge Nocedal
|
||||||
Copyright (c) 2007 Naoaki Okazaki
|
Copyright (c) 2007,2008 Naoaki Okazaki
|
||||||
|
|
||||||
Permission is hereby granted, free of charge, to any person obtaining a
|
Permission is hereby granted, free of charge, to any person obtaining a
|
||||||
copy of this software and associated documentation files (the "Software"),
|
copy of this software and associated documentation files (the "Software"),
|
||||||
|
56
ChangeLog
56
ChangeLog
@ -1,29 +1,53 @@
|
|||||||
2008-03-05 Naoaki Okazaki <okazaki at chokkan org>
|
2008-04-25 Naoaki Okazaki <okazaki at chokkan org>
|
||||||
|
|
||||||
* CRFsuite 0.4 (the first public release):
|
* libLBFGS 1.4:
|
||||||
- Website and documentation for CRFsuite.
|
- Configurable line search algorithms. A member variable
|
||||||
- Tutorial on the CoNLL 2000 chunking shared task.
|
lbfgs_parameter_t::linesearch was added to choose either MoreThuente
|
||||||
- Performance comparison on the CoNLL 2000 chunking shared task.
|
method (LBFGS_LINESEARCH_MORETHUENTE) or backtracking algorithm
|
||||||
- Bug fix in L2 regularization.
|
(LBFGS_LINESEARCH_BACKTRACKING).
|
||||||
- A number of small improvements for the public release.
|
- Fixed a bug: the previous version did not compute psuedo-gradients
|
||||||
|
properly in the line search routines for OW-LQN. This bug might quit
|
||||||
|
an iteration process too early when the OW-LQN routine was activated
|
||||||
|
(0 < lbfgs_parameter_t::orthantwise_c).
|
||||||
|
- Configure script for POSIX environments.
|
||||||
|
- SSE/SSE2 optimizations with GCC.
|
||||||
|
- New functions lbfgs_malloc and lbfgs_free to use SSE/SSE2 routines
|
||||||
|
transparently. It is uncessary to use these functions for libLBFGS
|
||||||
|
built without SSE/SSE2 routines; you can still use any memory
|
||||||
|
allocators if SSE/SSE2 routines are disabled in libLBFGS.
|
||||||
|
|
||||||
|
|
||||||
2007-12-12 Naoaki Okazaki <okazaki at chokkan org>
|
2007-12-16 Naoaki Okazaki <okazaki at chokkan org>
|
||||||
|
|
||||||
* CRFsuite 0.3 (internal release):
|
* libLBFGS 1.3:
|
||||||
- Implemented scaling method for forward/backward algorithm.
|
- An API change. An argument was added to lbfgs() function to receive
|
||||||
- Removed the code for computing the forward/backward algorithm in logarithm domain.
|
the final value of the objective function. This argument can be set
|
||||||
|
to NULL if the final value is unnecessary.
|
||||||
|
- Fixed a null-pointer bug in the sample code (reported by Takashi
|
||||||
|
Imamichi).
|
||||||
|
- Added build scripts for Microsoft Visual Studio 2005 and GCC.
|
||||||
|
- Added README file.
|
||||||
|
|
||||||
|
|
||||||
2007-11-30 Naoaki Okazaki <okazaki at chokkan org>
|
2007-12-13 Naoaki Okazaki <okazaki at chokkan org>
|
||||||
|
|
||||||
* CRFsuite 0.2 (internal release):
|
* libLBFGS 1.2:
|
||||||
- Orthant-Wise Limited-memory Quasi-Newton (OW-LQN) method for L1 regularization.
|
- Fixed a serious bug in orthant-wise L-BFGS. An important variable
|
||||||
|
was used without initialization.
|
||||||
- Configurable L-BFGS parameters (number of limited memories, epsilon).
|
- Configurable L-BFGS parameters (number of limited memories, epsilon).
|
||||||
|
|
||||||
|
|
||||||
2007-10-29 Naoaki Okazaki <okazaki at chokkan org>
|
2007-12-01 Naoaki Okazaki <okazaki at chokkan org>
|
||||||
|
|
||||||
* CRFsuite 0.1 (internal release):
|
* libLBFGS 1.1:
|
||||||
|
- Implemented orthant-wise L-BFGS.
|
||||||
|
- Implemented lbfgs_parameter_init() function.
|
||||||
|
- Fixed several bugs.
|
||||||
|
- API documentation.
|
||||||
|
|
||||||
|
|
||||||
|
2007-09-20 Naoaki Okazaki <okazaki at chokkan org>
|
||||||
|
|
||||||
|
* libLBFGS 1.0
|
||||||
- Initial release.
|
- Initial release.
|
||||||
|
|
||||||
|
@ -3,7 +3,7 @@
|
|||||||
SUBDIRS = lib sample
|
SUBDIRS = lib sample
|
||||||
|
|
||||||
docdir = $(prefix)/share/doc/@PACKAGE@
|
docdir = $(prefix)/share/doc/@PACKAGE@
|
||||||
doc_DATA = README INSTALL COPYING AUTHORS ChangeLog
|
doc_DATA = README INSTALL COPYING AUTHORS ChangeLog NEWS
|
||||||
|
|
||||||
EXTRA_DIST = \
|
EXTRA_DIST = \
|
||||||
autogen.sh \
|
autogen.sh \
|
||||||
|
14
README
14
README
@ -39,26 +39,26 @@ $ make install # To install libLBFGS library and header.
|
|||||||
=========================================================================
|
=========================================================================
|
||||||
This library has SSE/SSE2 optimization routines for vector arithmetic
|
This library has SSE/SSE2 optimization routines for vector arithmetic
|
||||||
operations on Intel/AMD processors. The SSE2 routine is for 64 bit double
|
operations on Intel/AMD processors. The SSE2 routine is for 64 bit double
|
||||||
values whereas the SSE routine is for 32 bit float values. Since the
|
values, and the SSE routine is for 32 bit float values. Since the default
|
||||||
default parameters for L-BFGS are tuned for double precision values, it
|
parameters in libLBFGS are tuned for double precision values, it may need
|
||||||
may be more straightforward to use SSE2 optimization than to use SSE.
|
to modify these parameters to use the SSE optimization routines.
|
||||||
|
|
||||||
To use the SSE2 optimization routine, specify --enable-sse2 option to the
|
To use the SSE2 optimization routine, specify --enable-sse2 option to the
|
||||||
configure script.
|
configure script.
|
||||||
|
|
||||||
$ ./configure --enable-sse2
|
$ ./configure --enable-sse2
|
||||||
|
|
||||||
To build libLBFGS with SSE optimization on Microsoft Visual Studio 2005,
|
To build libLBFGS with SSE2 optimization enabled on Microsoft Visual
|
||||||
define USE_SSE and __SSE2__ symbols.
|
Studio 2005, define USE_SSE and __SSE2__ symbols.
|
||||||
|
|
||||||
Make sure to run libLBFGS on processors where SSE2 instrunctions are
|
Make sure to run libLBFGS on processors where SSE2 instrunctions are
|
||||||
available.
|
available. The library does not check the existence of SSE2 instructions.
|
||||||
|
|
||||||
To package maintainers,
|
To package maintainers,
|
||||||
|
|
||||||
Please do not enable SSE/SSE2 optimization routine. The library built
|
Please do not enable SSE/SSE2 optimization routine. The library built
|
||||||
with SSE/SSE2 optimization will crash without any notice when necessary
|
with SSE/SSE2 optimization will crash without any notice when necessary
|
||||||
SSE/SSE2 instructions are unavailable.
|
SSE/SSE2 instructions are unavailable on computers.
|
||||||
|
|
||||||
|
|
||||||
|
|
||||||
|
Loading…
Reference in New Issue
Block a user