summaryrefslogtreecommitdiff
path: root/dtrain/README.md
diff options
context:
space:
mode:
authorPatrick Simianer <p@simianer.de>2011-10-14 15:40:23 +0200
committerPatrick Simianer <p@simianer.de>2011-10-14 15:40:23 +0200
commit628c4ecb641096c6526c7e6062460e627433f8fa (patch)
tree5cacd40d1bf27a18799eb2ab4c8b95cbfc986e58 /dtrain/README.md
parent73c406958feb6598382aee3b3043195e506a56f8 (diff)
test
Diffstat (limited to 'dtrain/README.md')
-rw-r--r--dtrain/README.md50
1 files changed, 50 insertions, 0 deletions
diff --git a/dtrain/README.md b/dtrain/README.md
new file mode 100644
index 00000000..dc980faf
--- /dev/null
+++ b/dtrain/README.md
@@ -0,0 +1,50 @@
+IDEAS
+=====
+ MULTIPARTITE ranking (108010, 1 vs all, cluster modelscore;score)
+ what about RESCORING?
+ REMEMBER kbest (merge) weights?
+ SELECT iteration with highest (real) BLEU?
+ GENERATED data? (multi-task, ability to learn, perfect translation in nbest, at first all modelscore 1)
+ CACHING (ngrams for scoring)
+ hadoop PIPES imlementation
+ SHARED LM (kenlm actually does this!)?
+ ITERATION variants
+ once -> average
+ shuffle resulting weights
+ weights AVERAGING in reducer (global Ngram counts)
+ BATCH implementation (no update after each Kbest list)
+ set REFERENCE for cdec (rescoring)?
+ MORE THAN ONE reference for BLEU?
+ kbest NICER (do not iterate twice)!? -> shared_ptr?
+ DO NOT USE Decoder::Decode (input caching as WordID)!?
+ sparse vector instead of vector<double> for weights in Decoder(::SetWeights)?
+ reactivate DTEST and tests
+ non deterministic, high variance, RANDOM RESTARTS
+ use separate TEST SET
+
+Uncertain, known bugs, problems
+===============================
+* cdec kbest vs 1best (no -k param), rescoring? => ok(?)
+* no sparse vector in decoder => ok/fixed
+* PhraseModel_* features (0..99 seem to be generated, why 99?)
+* flex scanner jams on malicious input, we could skip that
+
+FIXME
+=====
+* merge
+* ep data
+
+Data
+====
+<pre>
+nc-v6.de-en peg
+nc-v6.de-en.loo peg
+nc-v6.de-en.giza.loo peg
+nc-v6.de-en.symgiza.loo pe
+nv-v6.de-en.cs pe
+nc-v6.de-en.cs.loo pe
+--
+ep-v6.de-en.cs p
+ep-v6.de-en.cs.loo p
+</pre>
+