Loaded file ./tess4training-save/tess4training-vedic/tessdata/best/Devanagari.lstm, unpacking...
Warning: LSTMTrainer deserialized an LSTMRecognizer! Code range changed from 217 to 157!! Num (Extended) outputs,weights in Series: 1,48,0,1:1, 0 Num (Extended) outputs,weights in Series: C3,3:9, 0 Ft16:16, 160 Total weights = 160 [C3,3Ft16]:16, 160 Mp3,3:16, 0 Lfys64:64, 20736 Lfx64:64, 33024 Lrx64:64, 33024 Lfx512:512, 1181696 Fc157:157, 80541 Total weights = 1349181 Previous null char=2 mapped to 2 Continuing from ./tess4training-save/tess4training-vedic/tessdata/best/Devanagari.lstm See the line: Code range changed from 217 to 157!! That is the size of the unicharset. That is what is used in Fc. ShreeDevi ____________________________________________________________ भजन - कीर्तन - आरती @ http://bhajans.ramparivar.com On Wed, Aug 23, 2017 at 2:48 PM, <[email protected]> wrote: > Year, I have observed the builted network at beginning of the training > step. Thanks for reply. > > The basetrain.log file shows that Built network:[1,48,0,1 [C3,3 Ft16] > Mp3,3 Lfys64 Lfx96 Lrx96 Lfx512 Fc209] from request [1,48,0,1 Ct3,3,16 > Mp3,3 Lfys64 Lfx96 Lrx96 Lfx512 O1c1] > > Some problms for understanding this builted network: > > 1. [C3,3 Ft16] layers in the network has been enclosed in brackets. But > why it is enclosed in brackets? What does it stand for with the brackets? > 2. Fc209 the last layer of this network is a Fully-connected layer. > what's the meanings of the 'c' in this layer? I cannot find what 'Fc' > represents in the VGSLSpecs tutorial. > > Thanks. > > > > 在 2017年8月23日星期三 UTC+8下午3:00:00，shree写道： >> >> I think that number is ignored and the actual number generated from >> unichasrset is used. >> >> Usually there will be a message right at beginning of training showing >> the number being used. >> >> On 23-Aug-2017 12:21 PM, <[email protected]> wrote: >> >>> Hello, >>> >>> I have pulled out the network of the chi_sim.traineddata with the >>> command: combine_tessdata -u ../../tessdata/chi_sim.traineddata >>> ../../chi_sim_comp >>> >>> Then I observe the network which is shown in the chi_sim_comp file. The >>> network is [1,48,0,1Ct3,3,16 Mp3,3 Lfys64 Lfx96 Lrx96 Lfx512 O1c1] >>> >>> By analyzing the VGSL Specs language, I can infer that the output layer >>> of the network is O1c1, which means that Output layer produces 1-d >>> (sequence) output, trained with CTC,*outputting 1 class*. >>> >>> >>> Why does the output layer end up in one category? Whether the network >>> structure recorded in the chi_sim.traineddata will be wrong? >>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "tesseract-ocr" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to [email protected]. >>> To post to this group, send email to [email protected]. >>> Visit this group at https://groups.google.com/group/tesseract-ocr. >>> To view this discussion on the web visit https://groups.google.com/d/ms >>> gid/tesseract-ocr/5f5e3422-59e4-499e-bc4d-84ed214c1523%40goo >>> glegroups.com >>> <https://groups.google.com/d/msgid/tesseract-ocr/5f5e3422-59e4-499e-bc4d-84ed214c1523%40googlegroups.com?utm_medium=email&utm_source=footer> >>> . >>> For more options, visit https://groups.google.com/d/optout. >>> >> -- > You received this message because you are subscribed to the Google Groups > "tesseract-ocr" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at https://groups.google.com/group/tesseract-ocr. > To view this discussion on the web visit https://groups.google.com/d/ > msgid/tesseract-ocr/d0eec6d3-11af-4953-901a-4f5e03b63b79% > 40googlegroups.com > <https://groups.google.com/d/msgid/tesseract-ocr/d0eec6d3-11af-4953-901a-4f5e03b63b79%40googlegroups.com?utm_medium=email&utm_source=footer> > . > > For more options, visit https://groups.google.com/d/optout. > -- You received this message because you are subscribed to the Google Groups "tesseract-ocr" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/tesseract-ocr. To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/CAG2NduUjc2pTte%3DLWch_wNYjtA7qUyiCEzWQ%2BR3_CuySXWArKA%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
