From cslt Wiki
- LM count files still undelivered!
- Optimal Brain Damage(OBD).
- Online OBD held.
- OBD + L1 norm start to investigation.
- Graph search costs much. AM computing : graph search = 35: 65, for 1e-5 LM graph.
- Add AM score cache: AM computing + graph search = 25 : 75, reduced 5%.
- Investigating compact FST. Apply const FST first, and then try to implement LOUDS FST.
- Investigate spare computing
English NN training
- WSJ+Chinglish data training, done
- Prepare to test
- NN-based CSLM normalization. Normalize 10 NN networks by an additional hidden layer.
- 10240 X 10240 matrix, extremely slow. 5M data finished. PPL 289.
- Change the network to 10240 X 100 + 10240 X 100. 5M finished. PPL 241. 50M data: 224.
- 500 M running.
- Will start to implement the embedded stream mode.
- Database migration, done.
- Client bugs fixed.
- Recording scheduled.
- Text-based test completed. Correct answer/no answer/incorrect answer: 13:5:2