“2013-07-26”版本间的差异
来自cslt Wiki
（以内容“== Data sharing == * LM count files still undelivered! == DNN progress == === Experiments === * Sparse DNN on the ARM board <pre> 1200-1200-1200-3536 ...”创建新页面） |
|||
| 第31行： | 第31行： | ||
To be done: | To be done: | ||
| − | + | # SuiteSparse lib | |
| − | + | # decrease the hidden layer from 4-2 | |
| − | + | # Test accuracy on large data set | |
</pre> | </pre> | ||
| 第49行： | 第49行： | ||
To DO: | To DO: | ||
| − | + | # Large scale test. | |
| − | + | # CI Phone-based full path confidence estimation. | |
==FST sub-graph integration== | ==FST sub-graph integration== | ||
| − | + | # Coding finished. | |
| − | + | # Debuging and Test next week. | |
== Embedded progress == | == Embedded progress == | ||
| − | + | # Graph search costs a lot when the graph is large. Need to improve the indexing performance. | |
| − | + | # Need to integrate the Kaldi FE with pocket-sphinx decoder. | |
2013年7月26日 (五) 08:19的版本
目录
Data sharing
- LM count files still undelivered!
DNN progress
Experiments
- Sparse DNN on the ARM board
1200-1200-1200-3536 1200-1200-1200-3536-sparse0.3 (sparsity 1/5) original atlas: RT 2.3 RT 2.3 atlas sparse: RT 54 RT 14 NIST smatmat: RT 27.3 RT 5.98
800-800-800-2108 800-800-800-2108-sparse0.3 (sparsity 2/5): original atlas: RT 1.3 RT 1.1 NIST smatmat: RT 11.9 RT 5.5 600-600-600-1500 original atlas: RT 0.9 NIST smatmat: RT 6.5
To be done: # SuiteSparse lib # decrease the hidden layer from 4-2 # Test accuracy on large data set
Tencent exps
GPU & CPU merge
- Hold
Confidence estimation
1. DNN-based confidence estimation for decoding and alignment is done. The intuitive testing seems ok.
To DO:
- Large scale test.
- CI Phone-based full path confidence estimation.
FST sub-graph integration
- Coding finished.
- Debuging and Test next week.
Embedded progress
- Graph search costs a lot when the graph is large. Need to improve the indexing performance.
- Need to integrate the Kaldi FE with pocket-sphinx decoder.
