diff options
| author | Soumith Chintala <soumith@gmail.com> | 2015-10-14 17:41:21 +0300 |
|---|---|---|
| committer | Soumith Chintala <soumith@gmail.com> | 2015-10-14 17:41:21 +0300 |
| commit | 37b1e723874fa1ab8a5fb26519e2e9a2cd8dbd7e (patch) | |
| tree | a66a69b36793d0e74618a57710fc6eab714842a6 | |
| parent | 4da9f2440701d71b01865a1df7fe63177226c9da (diff) | |
| parent | 489cf76bf00abe7752a0ea4ed518d285d50e8fc8 (diff) | |
Merge pull request #31 from Moodstocks/doc-serialize
README: fix shared serialization package name
| -rw-r--r-- | README.md | 8 |
1 files changed, 4 insertions, 4 deletions
@@ -277,7 +277,7 @@ This method will call [synchronize](#threads.synchronize), terminate each queue <a name='threads.serialization'/> #### Threads.serialization(pkgname) #### -Specify which serialization scheme which should be used. This function +Specify which serialization scheme should be used. This function should be called (if you want a particular serialization) before calling [threads.Threads()](#threads.Threads) constructor. @@ -285,11 +285,11 @@ A serialization package (`pkgname`) should return a table of serialization functions when required (`save` and `load`). See [serialize specifications](#threads.serialize) for more details. -By default the serialization system uses the `threads.serialize` sub-package, which leverages torch serialization. +By default the serialization system uses the `'threads.serialize'` sub-package, which leverages torch serialization. -The `threads.sharedserialization` is also provided, which transparently +The `'threads.sharedserialize'` sub-package is also provided, which transparently *shares* the storages, tensors and [tds](http://github.com/torch/tds) C -data structures. This approach is great if one need to pass large data +data structures. This approach is great if one needs to pass large data structures between threads. See [the shared example](test/test-threads-shared.lua) for more details. |