<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"> <head> <meta http-equiv="Content-Type" content="text/html; charset=us-ascii"> <meta name="Generator" content="Microsoft Word 14 (filtered medium)"> <style></style> </head> <body lang="EN-US" link="blue" vlink="purple"> <div class="WordSection1"> Yes, Woodley is right, the problem is with our own (Hebrew) grammar.<o:p></o:p> We tried generating with the ERG just to make sure that it's not a more general problem.<o:p></o:p> <o:p> </o:p> We'll try the –vv option and see what we find.<o:p></o:p> In any case, Woodley, we'll gladly email you the grammar in the next few days.<o:p></o:p> <o:p> </o:p> Thanks!<o:p></o:p> <o:p> </o:p> <div> <div style="border:none;border-top:solid #B5C4DF 1.0pt;padding:3.0pt 0cm 0cm 0cm"> From: Woodley Packard [mailto:sweaglesw@sweaglesw.org] Sent: Thursday, February 05, 2015 8:36 PM To: Emily M. Bender Cc: Nurit Melnik; developers@delph-in.net Subject: Re: [developers] Generation in ACE on Ubuntu+LKB<o:p></o:p> </div> </div> <o:p> </o:p> <div> Hi,<o:p></o:p> </div> <div> <o:p> </o:p> </div> It sounds to me like Nurit is having trouble generating with his own (English) grammar, not with the ERG.<o:p></o:p> <div> <o:p> </o:p> </div> <div> It's surprising to see an out-of-memory condition for such a short sentence.  I recommend passing -vv to the generator.  You will see a complete printout of all of the edges (partial analyses) that are generated.  You may discover something surprising.  It will probably produce a lot of output, so piping it to a file will be helpful.  The difference vs. the LKB is surprising, so I would be interested in reproducing the problem locally if you are willing to share the grammar files.<o:p></o:p> </div> <div> <o:p> </o:p> </div> <div> Unfortunately, not all of the LUI menu callbacks are implemented in ACE's LUI driving mode, so the errors about browsing edge 6's "generate" are expected.  The pipeline approach you took was the right one.<o:p></o:p> </div> <div> <o:p> </o:p> </div> <div> Best of luck,<o:p></o:p> </div> <div> Woodley Packard<o:p></o:p> </div> <div> <o:p> </o:p> <div> <div> On Feb 5, 2015, at 8:53 AM, "Emily M. Bender" <<a href="mailto:ebender@uw.edu">ebender@uw.edu</a>> wrote:<o:p></o:p> </div> <o:p></o:p> <div> <div> <div> Hi Nurit,<o:p></o:p> </div> We're using Ubuntu+LKB in 567 this quarter, but not making use of the ERG, so I don't have a lot of info for you.  In case it helps, though:<o:p></o:p> </div> (NB: This is with the test version of Ubuntu+LKB, from just before the latest release.) Emily<o:p></o:p> <div> ubuntu@UbuntuLKB:~/erg$ echo "Dan ate" | ace -g erg-1212-x86-64-0.9.17.dat NOTE: 0 readings, added 311 / 26 edges to chart (16 fully instantiated, 15 actives used, 7 passives used)    RAM: 638k SKIP: Dan ate NOTE: parsed 0 / 1 sentences, avg 1103k, time 0.00671s ubuntu@UbuntuLKB:~/erg$ echo "The dog barks." | ace -g erg-1212-x86-64-0.9.17.dat -T 2>/dev/null SENT: The dog barks. [ LTOP: h0 INDEX: e2 [ e SF: prop TENSE: pres MOOD: indicative PROG: - PERF: - ] RELS: < [ _the_q_rel<0:3> LBL: h4 ARG0: x3 [ x PERS: 3 NUM: sg IND: + ] RSTR: h5 BODY: h6 ]  [ "_dog_n_1_rel"<4:7> LBL: h7 ARG0: x3 ]  [ "_bark_v_1_rel"<8:14> LBL: h1 ARG0: e2 ARG1: x3 ] > HCONS: < h0 qeq h1 h5 qeq h7 > ] [ LTOP: h0 INDEX: e2 [ e SF: prop ] RELS: < [ unknown_rel<0:14> LBL: h1 ARG0: e2 ARG: x4 [ x PERS: 3 NUM: pl ] ]  [ _the_q_rel<0:3> LBL: h5 ARG0: x4 RSTR: h6 BODY: h7 ]  [ compound_rel<4:14> LBL: h8 ARG0: e9 [ e SF: prop TENSE: untensed MOOD: indicative PROG: - PERF: - ] ARG1: x4 ARG2: x10 [ x IND: + ] ]  [ udef_q_rel<4:7> LBL: h11 ARG0: x10 RSTR: h12 BODY: h13 ]  [ "_dog_n_1_rel"<4:7> LBL: h14 ARG0: x10 ]  [ "_bark_n_1_rel"<8:14> LBL: h8 ARG0: x4 ] > HCONS: < h0 qeq h1 h6 qeq h8 h12 qeq h14 > ] ubuntu@UbuntuLKB:~/erg$ echo "The dog barks." | ace -g erg-1212-x86-64-0.9.17.dat -T 2>/dev/null | ace -g erg-1212-x86-64-0.9.17.dat -e The dog barks. NOTE: 62 passive, 143 active edges in final generation chart; built 65 passives total. [1 results] The dog barks. NOTE: 53 passive, 241 active edges in final generation chart; built 57 passives total. [1 results] NOTE: generated 2 / 2 sentences, avg 1569k, time 0.08293s NOTE: transfer did 409 successful unifies and 393 failed ones ubuntu@UbuntuLKB:~/erg$ echo "The dog eats." | ace -g erg-1212-x86-64-0.9.17.dat -T 2>/dev/null | ace -g erg-1212-x86-64-0.9.17.dat -e The dog eats. NOTE: 90 passive, 157 active edges in final generation chart; built 93 passives total. [1 results] The dog eats. NOTE: 49 passive, 227 active edges in final generation chart; built 53 passives total. [1 results] NOTE: generated 2 / 2 sentences, avg 1674k, time 0.02566s NOTE: transfer did 409 successful unifies and 393 failed ones ubuntu@UbuntuLKB:~/erg$ echo "The dog eats cake." | ace -g erg-1212-x86-64-0.9.17.dat -T 2>/dev/null | ace -g erg-1212-x86-64-0.9.17.dat -e The dog eats cake. NOTE: 116 passive, 190 active edges in final generation chart; built 127 passives total. [1 results] The dog eats cake. NOTE: 156 passive, 253 active edges in final generation chart; built 163 passives total. [1 results] The dog, which, eats cake. The dog, that, eats cake. The dog, who, eats cake. The dog, which eats cake. The dog, who eats cake. The dog, that eats cake. The dog which, eats cake. The dog that, eats cake. The dog who, eats cake. The dog which eats cake. The dog that eats cake. The dog who eats cake. The dog eats cake. NOTE: 84 passive, 227 active edges in final generation chart; built 111 passives total. [13 results] NOTE: generated 3 / 3 sentences, avg 2255k, time 0.13414s NOTE: transfer did 996 successful unifies and 837 failed ones<o:p></o:p> </div> </div> <div> <o:p> </o:p> <div> On Thu, Feb 5, 2015 at 5:21 AM, Nurit Melnik <<a href="mailto:nuritme@openu.ac.il" target="_blank">nuritme@openu.ac.il</a>> wrote:<o:p></o:p> <div> <div> Hello,<o:p></o:p>  <o:p></o:p> I've recently started working with ACE on the Ubuntu+LKB platform (version 21, 64-bit).<o:p></o:p>  <o:p></o:p> With the LKB I managed to generate all input with my grammar, by using the right-click option on the parse tree.<o:p></o:p> With ACE I can generate with ERG, but I have problems with my grammar.<o:p></o:p>  <o:p></o:p> Intransitives work fine:<o:p></o:p> echo "Dan ate" | ace -g grammar.dat  -T 2>/dev/null | ace -g grammar.dat -e <o:p></o:p> But with “Dan ate cake” I get the following error: <o:p></o:p>  <o:p></o:p> "NOTE: terminating search, too much RAM"<o:p></o:p>  <o:p></o:p> I tried  generating by using the right-click option on the parse tree (in LUI mode).<o:p></o:p> I got the following message with ERG and with my grammar, regardless of valence:<o:p></o:p>  <o:p></o:p> "LUI: don't know how to brows edge 6's 'generate' " (where “6” depends on the sentence.)<o:p></o:p>  <o:p></o:p> This also happens with other right-click options, except for AVMs and simple MRS.<o:p></o:p>  <o:p></o:p> What am I missing here?<o:p></o:p>  <o:p></o:p> Thanks in advance,<o:p></o:p> Nurit<o:p></o:p>  <o:p></o:p> </div> </div> </div> -- <o:p></o:p> <div> <div> Emily M. Bender Professor, Department of Linguistics Check out CLMS on facebook! <a href="http://www.facebook.com/uwclma" target="_blank"> http://www.facebook.com/uwclma</a><o:p></o:p> </div> </div> </div> </div> <o:p> </o:p> </div> </div> </body> </html>