|
SPRAAK
|
All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Groups Pages
|
SPRAAK
|
Collaboration diagram for SprCwrLMStd:Data Fields | |
| char | allocated |
| this structure is allocated, and hence must be deallocated More... | |
| char | check_lvl |
| check the LM and its compatibility with the lexicon More... | |
| char | bo0_elmc |
| set the back-off cost of empty LM-contexts to zero More... | |
| char | slmsp_method |
| method used to calculate static LM-smearing probs More... | |
| int | eos_ndx |
| int | N |
| the N in N-gram More... | |
| int | Nword |
| the number of words in the LM More... | |
| ssize_t | Nlmc |
| the number of distinct LM-contexts (prob. distributions) More... | |
| ssize_t | Nprob |
| the number of probabilities More... | |
| ssize_t | sz_lmc_data |
| just for statistics More... | |
| int | sz_wstr |
| just for statistics More... | |
| int | sz_qprob |
| just for statistics More... | |
| int | sz_qsize |
| just for statistics More... | |
| int * | qprob |
| the quantized probabilities More... | |
| SprCwrLMStd32SQ * | qsize |
| quantized size combinations (Ntot,N32) More... | |
| unsigned short * | lmc_data |
| the word lists with their probs for each LM-context More... | |
| unsigned short * | lmc_prob |
| unsigned short ** | lmc_ptr |
| unsigned short * | lmc_disc0 |
| unsigned int | sz_hash_tbl |
| size of the LM-context hash table More... | |
| char * | wstr |
| all words in a single large string block More... | |
| SprStrHashTbl | wlst |
| hash table used to convert the words into numbers (indices) More... | |
| double | prob_sf |
| SprCwrLMExt * | ext |
| the LM-extensions (optional) More... | |
| char SprCwrLMStd::allocated |
this structure is allocated, and hence must be deallocated
| char SprCwrLMStd::check_lvl |
check the LM and its compatibility with the lexicon
| char SprCwrLMStd::bo0_elmc |
set the back-off cost of empty LM-contexts to zero
| char SprCwrLMStd::slmsp_method |
method used to calculate static LM-smearing probs
| int SprCwrLMStd::eos_ndx |
index of the end of sentence word (can't go beyond this word without some help/extensions)
| int SprCwrLMStd::N |
the N in N-gram
| int SprCwrLMStd::Nword |
the number of words in the LM
| ssize_t SprCwrLMStd::Nlmc |
the number of distinct LM-contexts (prob. distributions)
| ssize_t SprCwrLMStd::Nprob |
the number of probabilities
| ssize_t SprCwrLMStd::sz_lmc_data |
just for statistics
| int SprCwrLMStd::sz_wstr |
just for statistics
| int SprCwrLMStd::sz_qprob |
just for statistics
| int SprCwrLMStd::sz_qsize |
just for statistics
| int* SprCwrLMStd::qprob |
the quantized probabilities
| SprCwrLMStd32SQ* SprCwrLMStd::qsize |
quantized size combinations (Ntot,N32)
32-bits LM's only, NULL for 16-bits LM's
| unsigned short* SprCwrLMStd::lmc_data |
the word lists with their probs for each LM-context
| unsigned short* SprCwrLMStd::lmc_prob |
the prob value for children that are LM-contexts in itself, accessed through a hashing function
| unsigned short** SprCwrLMStd::lmc_ptr |
indices to the LM-context blocks in lmc_data, accessed through a hashing function + subsampled at 1/8th (jumping to the next is done simply by adding the sizes of the lmc_blocks)
| unsigned short* SprCwrLMStd::lmc_disc0 |
pointer to the dummy empty context, used when the discount tree countains holes
| unsigned int SprCwrLMStd::sz_hash_tbl |
size of the LM-context hash table
| char* SprCwrLMStd::wstr |
all words in a single large string block
| SprStrHashTbl SprCwrLMStd::wlst |
hash table used to convert the words into numbers (indices)
| double SprCwrLMStd::prob_sf |
scale factor that must be used to obtain log(probs) (natural log)
| SprCwrLMExt* SprCwrLMStd::ext |
the LM-extensions (optional)
1.8.6
