Infrastructure/software/frameworks - Revision history

Oe: /* Background */

2019-05-11T22:14:01Z

‎Background

2019-05-11T22:12:11Z

‎Background

2018-10-08T19:50:16Z

‎Background

2018-09-29T23:02:21Z

‎Background

2018-09-29T20:51:08Z

‎Background

2018-09-29T20:44:22Z

‎Background

2018-09-19T14:24:03Z

2018-03-06T15:00:40Z

‎Vision

2018-03-06T14:43:56Z

‎Vision

2018-03-06T14:39:07Z

‎Vision

@@ Line 28: / Line 28: @@
 module use -a /proj*/nlpl/software/modulefiles
 module load nlpl-nltk/3.4/3.5 nlpl-pytorch/0.4.1 nlpl-tensorflow/1.11
-cat << EOF | python
+cat << EOF | python3
 import nltk; import torch; import tensorflow;
 print(nltk.__version__);

@@ Line 27: / Line 27: @@
 module purge
 module use -a /proj*/nlpl/software/modulefiles
-module load nlpl-nltk nlpl-pytorch nlpl-tensorflow
+module load nlpl-nltk/3.4/3.5 nlpl-pytorch/0.4.1 nlpl-tensorflow/1.11
 cat << EOF | python
 import nltk; import torch; import tensorflow;

@@ Line 26: / Line 26: @@
 <pre>
 module purge
-module use -a /projects/nlpl/software/modulefiles
+module use -a /proj*/nlpl/software/modulefiles
 module load nlpl-nltk nlpl-pytorch nlpl-tensorflow
 cat << EOF | python

@@ Line 16: / Line 16: @@
 In mid-2018, the NLPL infrastructure task force has at least temporarily
 deviated from some of the above and has created project-owned installations
-of PyTorch (on Abel and Taito) and TensorFlow (so far, on Abel only).
+of PyTorch and TensorFlow (on both systems).
 There are certain benefits to this approach too, notably increased
 parallelism across the two systems and tighter integration with other

@@ Line 19: / Line 19: @@
 There are certain benefits to this approach too, notably increased
 parallelism across the two systems and tighter integration with other
-elements of the project-specific NLPL installations.
+elements of the project-specific NLPL installations, for example
 = Taito =

@@ Line 4: / Line 4: @@
 are a prerequisite to much current NLP research, somewhat analogous to
 compilers or core libraries (like Boost or ICU).
-These frameworks are not in principle discipline-specific and will often
+These frameworks are not in principle discipline-specific and will sometimes
 be difficult to install in user space (i.e. without administrator privileges).
-Thus, general DL support should be provided at the system level (not on a
+Thus, general DL support should arguably be provided at the system level (not on a
-per-project basis), and NLPL will seek to work with the system administrators
+per-project basis), and NLPL seeks to work with the system administrators
 to make sure that (a) relevant frameworks are available (in all relevant
 versions, both for CPU and GPU usage) and (b) that system-wide installations
@@ Line 13: / Line 13: @@
 to NLPL users.
 These are no small ambitions :-).
 = Taito =

← Older revision	Revision as of 14:24, 19 September 2018
(No difference)

@@ Line 45: / Line 45: @@
 (a) availability and support for all relevant DL frameworks,
 in (b) multiple, clearly discernible release versions as the
-frameworks involve.
+frameworks evolve.
 It should be possible to (c) combine multiple frameworks with
 each other (the Uppsala parser, for example, internally

@@ Line 44: / Line 44: @@
 The ideal situation, from the NLPL point of view, would be
 (a) availability and support for all relevant DL frameworks,
-in (b) multiple, clearly discernible versions as the
+in (b) multiple, clearly discernible release versions as the
 frameworks involve.
 It should be possible to (c) combine multiple frameworks with
@@ Line 53: / Line 53: @@
 be (d) uniform to the highest degree possible across the
 two systems.
 How to best accomplish (something resembling) the above is
 no trivial question.
-Possibly the ‘bundle’ approach pursued at CSC (in their
+Possibly the [https://research.csc.fi/-/mlpython ‘bundle’]

← Older revision		Revision as of 14:39, 6 March 2018
Line 53:		Line 53:
	be (d) uniform to the highest degree possible across the		be (d) uniform to the highest degree possible across the
	two systems.		two systems.
		+
		+	How to best accomplish (something resembling) the above is
		+	no trivial question.
		+	Possibly the ‘bundle’ approach pursued at CSC (in their