Searched refs:grams (Results 1 – 14 of 14) sorted by relevance
117 skip grams. The configurable parameters are `ngram_size` and123 This is a custom op that hashes skip grams to features represented as124 integers. Longer skip-grams are allocated higher weights.
181 // A flat list of all the hashed n-grams concatenated back to back. Elements185 // Offsets to the start of the n-grams in hashed_ngram_tokens. The last190 // Weights of the n-grams.193 // The default weight assigned to n-grams that weren't matched.198 // if additional (longer) n-grams are added to a model as part of a minor
198 getbaseclasses(grams)393 'gram' : grams,
1010 class grams(aetools.ComponentItem): class1614 grams._superclassnames = []1615 grams._privpropdict = {1617 grams._privelemdict = {2099 'gram' : grams,
474 dnam{"grams"}476 other{"{0} grams"}1631 dnam{"grams"}
479 dnam{"grams"}481 other{"{0} grams"}1968 dnam{"grams"}
569 one{"{0} grams"}
261 cryp-to-grams
262 cryp-to-grams
707 <td>Specifies the media weight in grams per meter<sup>2</sup>.</td>
4789 grams that need the best possible performance, there is also a "fast
69359 grams %36609 græmz
82265 grams %25815 græmz